Modeling the impacts of land-use change on ecosystems at the regional and continental scale
Jennifer Koch
ISBN 978-3-89958-964-1
Jennifer Koch
Modeling the impacts of land-use
change on ecosystems at the
regional and continental scale
Jennifer Koch
Modeling the impacts of land-use
change on ecosystems at the
regional and continental scale
kassel
university
press
This work has been accepted by the faculty of Electrical Engineering and Computer Science of the
University of Kassel as a thesis for acquiring the academic degree of Doktorin der Ingenieurwissenschaften
(Dr.-Ing.).
Supervisor:
Co-Supervisor:
Prof. Dr. Joseph Alcamo, CESR, Universität Kassel
Prof. Dr. Gerhard Gerold, Georg-August-Universität Göttingen
Defense day:
Bibliographic information published by Deutsche Nationalbibliothek
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data is available in the Internet at http://dnb.d-nb.de.
Zugl.: Kassel, Univ., Diss. 2010
ISBN print: 978-3-89958-964-1
ISBN online: 978-3-89958-965-8
URN: http://nbn-resolving.de/urn:nbn:de:0002-9656
© 2010, kassel university press GmbH, Kassel
www.upress.uni-kassel.de
Printed in Germany
21 June 2010
st
Danksagung
Die vorliegende Arbeit entstand ihm Rahmen des Projektes “GLOWA Jordan River” am
Center for Environmental Systems Research (CESR) der Universität Kassel. Die Finanzierung erfolgte durch das Bundesministerium für Bildung und Forschung.
Mein herzlicher Dank gilt meinem Doktorvater Prof. Dr. Joseph Alcamo für die weitsichtige wissenschaftliche Beratung und Unterstützung. Außerdem möchte ich mich herzlich
bei Prof. Dr. Gerold für die Begutachtung dieser Arbeit bedanken. Ein sehr herzlicher Dank
geht an Dr.-Ing. Rüdiger Schaldach für die hevorragende fachliche Betreuung dieser Arbeit
und für seine einmalige Fähigkeit zu motivieren. Weiterhin gilt mein Dank den Kolleginnen
und Kollegen am CESR für die angenehme Arbeitsatmosphäre und die schöne Zeit. Besonders bedanken möchte ich mich bei Dr.-Ing. Martina Weiß und Dr. Janina Onigkeit für
die vielen konstruktiven Gespräche, sowie bei Achim Manche für die fundierte technische
Beratung.
Nicht zuletzt danke ich meiner Familie und besonders meinen Eltern für ihre ständige
Förderung und Unterstützung!
iii
Zusammenfassung
Die Ausdehnung landwirtschaftlich genutzter Flächen und die Intensivierung der Landwirtschaft können zu Veränderungen der Stabilität und Struktur von Ökosystem führen.
Diese Veränderungen können das Vermögen von Ökosystemen, Dienstleistungen zu erbringen, reduzieren. Prognosen lassen einen Anstieg der Weltbevölkerung um 2.2 Milliarden
Menschen während der nächsten vierzig Jahre erwarten. Mit diesem Anstieg geht vermutlich eine Verdopplung des weltweiten Nahrungsmittelbedarfs bis zum Jahr 2050 einher.
Um diesen steigenden Bedarf zu erfüllen, sind zusätzliche landwirtschaftliche Aktivitäten
notwendig. Diese haben eine weitere Beanspruchung von Ökosystemen zur Folge. Dadurch
wird es unerlässlich ökologisch angepasste Bewirtschaftungsstrategien zu entwickeln, die
es ermöglichen Ökosystemfunktionen zu erhalten und gleichzeitig den Nahrungsmittelbedarf zu decken. Eine Möglichkeit, die Ausarbeitung solcher Bewirtschaftungsstrategien zu
unterstützen, ist die Anwendung von Landnutzungsmodellen in Form von Szenariostudien.
In der vorliegenden Arbeit wurden ökologisch basierte Bewirtschaftungsstrategien mit
Hilfe des integrativen Landnutzungsmodells LandSHIFT untersucht. Der Schwerpunkt lag
dabei auf der Bewirtschaftung von Weideland in der Jordanregion und auf der Ausweisung von Naturschutzgebieten in Afrika. Für die Untersuchung der Bewirtschaftung von
Weideland in der Jordanregion wurde eine regionale LandSHIFT-Version (LandSHIFT.R)
entwickelt. LandSHIFT.R bildet einen Rückkopplungsmechanismus zwischen Viehbesatzdichte und Biomasseproduktivität ab und ermöglicht somit die Untersuchung verschieden
intensiver Beweidungsstrategien auf Grundlage von Besatzkapazitäten. Zur Untersuchung
der Ausweisung von Naturschutzgebieten in Afrika wurde die bereits existierende, globale LandSHIFT-Version um einen Algorithmus erweitert, der es ermöglicht, eingeschränkte
Landnutzungsaktivitäten in Regionen mit hoher Gefäßpflanzenvielfalt zu simulieren.
Um die Funktionalitäten von LandSHIFT.R zu testen, wurde ein erstes Simulationsexperiment durchgeführt. Basierend auf verschiedenen Annahmen zur Bewirtschaftung von
Weideland und zu Landdegradation aufgrund von Überweidung, wurden mögliche zukünftige Entwicklungen von Landnutzung und Landbedeckung analysiert. Die Ergebnisse des
Simulationsexperiments zeigen, dass die Entwicklung von Landnutzung und Landbedeckung sensibel auf den implementierten Rückkopplungsmechanismus sowie die verschiedenen Strategien zur Bewirtschaftung von Weideland reagiert.
Anschließend wurden mit LandSHIFT.R zwei Simulationen durchgeführt, die sich ausschließlich in der zugrunde liegenden Strategie zur Bewirtschaftung von Weideland unterscheiden. Um die Auswirkungen der verschiedenen Strategien auf Biomasseproduktivität
und Landschaftsstrukturen zu quantifizieren, wurden die Simulationsergebnisse mit einem
Indikator zur Bewertung der Biodiversitätsbelastung sowie mit verschiedenen Landschaftsmetriken ausgewertet. Die Ergebnisse zeigen einen deutlich höheren Bedarf an Weidefläche unter nachhaltiger Bewirtschaftung verglichen mit dem Bedarf an Weidefläche bei
intensiver Bewirtschaftung. Die Auswertung der Besatzdichten zeigt jedoch eine höhere
Biodiversitätsbelastung bei intensiver Bewirtschaftung von Weideland. Zusätzlich zeigt die
v
Auswertung mit einer Kombination der o.g. Indikatoren eine deutlichere Fragmentierung
der Landschaft bei intensiver Bewirtschaftung von Weideland.
Um mögliche zukünftige Entwicklungspfade für die Jordanregion zu untersuchen, wurde
LandSHIFT.R validiert und anschließend im Rahmen einer Szenarienanalyse eingesetzt. Insgesamt wurden vier Simulationen durchgeführt, die sich hinsichtlich der sozioökonomischen
Treiber von Landnutzungsänderungen und der Strategie zur Bewirtschaftung von Weideland unterschieden. Die Ergebnisse der Validierung zeigen, dass LandSHIFT.R vornehmlich
dort Landnutzungsänderungen ausweist, wo auch tatsächlich Änderungen beobachtet wurden. Die Ergebnisse der Simulationen lassen darauf schließen, dass das optimistischste
Szenario für die Menschen in der Jordanregion, trotz intensiver Bewirtschaftung von Weideland, die stärkste Ausdehnung von landwirtschaftlich genutzter Fläche mit sich bringt
und somit zu der stärksten Belastung der regionalen Ökosysteme führt.
Mit Hilfe der erweiterten, globalen LandSHIFT-Version wurden die Flächenpotenziale für
Feldfruchtproduktion und Beweidung in Afrika quantifiziert. Des Weiteren wurde die mögliche Bedrohung der Artenvielfalt von Gefäßpflanzen durch zukünftige landwirtschaftliche
Entwicklung analysiert. Die Untersuchung wurde separat für die Landwirtschaftsformen
Regenfeldbau, Bewässerungsfeldbau und Weidewirtschaft durchgeführt. Die Analyse ergab Flächenpotenziale von circa 2 000 Millionen Hektar für Regenfeldbau, 2 800 Millionen
Hektar für Bewässerungsfeldbau und 1 200 für Weidewirtschaft. Außerdem zeigte sich eine deutliche Überschneidung der Flächenpotenziale für landwirtschaftliche Aktivitäten mit
Regionen, die eine hohe Vielfalt an Gefäßpflanzen aufweisen. Dies weist auf den hohen
Stellenwert großräumiger Naturschutzstrategien zur Erhaltung von Ökosystemfunktionen
hin. Die Einschränkung von Landnutzungsaktivitäten in Regionen mit hoher Gefäßpflanzendiversität bewirkte keine starke Zunahme von zusätzlichen Landnutzungsänderungen. Dies
zeigt, dass die Ausweisung neuer Naturschutzgebiete in Regionen mit hoher Artenvielfalt
wahrscheinlich keine wesentliche Beeinträchtigung der landwirtschaftlichen Entwicklung in
Afrika darstellt.
vi
Summary
Agricultural intensification and the expansion of agricultural area can induce changes in
structure and resilience of ecosystems and thus reduce the capacity of ecosystems to provide goods and services in the long run. Over the next forty years, the world population is
expected to increase by 2.2 billion. This is likely to result in a doubling of the world food
demand by 2050. Agricultural activities, essential to meet this food demand, will inevitably
put additional pressure on ecosystems. Hence, there is an urgent need for ecologically based management strategies that help to meet human demands and maintain the provision
of ecosystem functions for future generations. The application of simulation models of
land-use change in the context of scenario studies can support the development of such
strategies.
In this thesis, the integrated land-use change model LandSHIFT was applied in order
to investigate ecologically based agricultural management strategies, focusing on rangeland management in the Jordan River region and nature conservation in Africa. For the
Jordan River study, a regional-scale LandSHIFT version (LandSHIFT.R) was constructed,
which incorporates the feedback between stocking density and biomass productivity and
the restriction of stocking densities based on stocking capacities. Furthermore, the globalscale LandSHIFT version was extended by an algorithm that allows to simulate gradual
constraint of land-use activities in African regions with high vascular plant diversity.
To test the functionalities of LandSHIFT.R, an initial simulation experiment was conducted. Based on various assumptions on rangeland management strategies and on landscape
degradation due to overgrazing, the possible future development of land-use and land-cover
change was assessed. The results prove that the computed land-use and land-cover changes are sensitive to the implemented feedback mechanism and to the different rangeland
management strategies.
LandSHIFT.R was then applied in a second simulation experiment to carry out two
simulations, varying exclusively in the applied rangeland management strategy. The results
of these simulations were evaluated with an indicator for pressure on biodiversity and with
a set of landscape metrics, in order to quantify the effect of different grazing intensities
on biomass productivity and landscape structure. The results show that the rangeland
area demand under sustainable rangeland management exceeds the one under intensive
rangeland management by far. However, the evaluation of the stocking densities indicates
a higher pressure on biodiversity under intensive rangeland management. Moreover, the
evaluation of rangeland area with a combination of landscape pattern metrics and the
indicator for pressure on biodiversity revealed a stronger fragmentation of the landscape
under intensive rangeland management.
To investigate potential future development pathways for the Jordan River region, LandSHIFT.R was validated and then applied in the context of an environmental scenario analysis. Altogether, four simulations were conducted, varying in socio-economic drivers and
rangeland management strategy. The validation results show that LandSHIFT.R prefe-
vii
rentially simulates changes in land-use and land-cover at locations, where land-use and
land-cover changes were actually observed. The results of the simulations show that the
most optimistic scenario for the people in the Jordan River region results in the strongest
expansion of agricultural area and, hence, is the one that is likely to put the highest pressure
on the natural resources.
The revised global-scale LandSHIFT version was applied for the quantification of the
area potentials for crop production and livestock grazing in Africa. Furthermore, the possible threat of future agricultural development to biodiversity was assessed. The analysis
was performed separately for the agricultural activities rain-fed crop production, irrigated
crop production, and livestock grazing. The simulation results indicate that there is an area
potential of about 2 000 million hectare for rain-fed crop production, 2 800 million hectare
for irrigated crop production, and 1 200 million hectare for livestock grazing. The analysis
reveals that there is a significant overlap of area potentials for agricultural activities and
regions with high vascular plant diversity. This indicates a high importance of large-scale
conservation strategies in order to maintain ecosystem functions. The limitation of agricultural activities in regions with high vascular plant diversity did not result in a strong
increase of land conversion. Hence, from the continental perspective, the implementation
of new protection areas, based on regions with high vascular plant diversity, is not likely to
substantially hinder agricultural development in Africa.
viii
Table of contents
List of figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
List of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
1. Introduction
1.1. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2. Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.3. Structure of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
1
2
3
2. Land-use and land-cover change: terminology and concepts
2.1. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2. Causes and consequences of land-use and land-cover change . . . . . . . .
2.3. Main approaches to modeling land-use and land-cover change . . . . . . .
5
5
6
7
3. The integrated modeling system LandSHIFT
3.1. Overview . . . . . . . . . . . . . . . . . . . .
3.1.1. Purpose . . . . . . . . . . . . . . . .
3.1.2. State variables and scales . . . . . . .
3.1.3. Process overview and scheduling . . .
3.2. Design concepts . . . . . . . . . . . . . . . .
3.3. Details . . . . . . . . . . . . . . . . . . . . .
3.3.1. Initialization . . . . . . . . . . . . . .
3.3.2. Input . . . . . . . . . . . . . . . . . .
3.3.3. Sub models . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4. Modeling the feedback between stocking density and biomass productivity
4.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2. Simulation experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3. Study region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4. Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.1. Input data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.2. Scenario description . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.3. Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5.1. Simulation of land-use and land-cover change . . . . . . . . . . . .
11
11
11
12
14
16
18
18
19
19
33
33
34
35
36
36
38
38
39
41
ix
Table of contents
4.5.2. Validation results . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.6. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5. Quantifying the environmental impact of
5.1. Motivation . . . . . . . . . . . . . . . .
5.2. Simulation experiment . . . . . . . . . .
5.3. Study region . . . . . . . . . . . . . . .
5.4. Materials and methods . . . . . . . . . .
5.4.1. Relative HANPP . . . . . . . . .
5.4.2. Landscape metrics . . . . . . . .
5.4.3. Validation . . . . . . . . . . . .
5.5. Results . . . . . . . . . . . . . . . . . .
5.6. Discussion and conclusions . . . . . . . .
grazing in Jordan
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . .
6. Future land-use and land-cover change scenarios for the
region
6.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2. Study region . . . . . . . . . . . . . . . . . . . . . . . . .
6.3. Materials and methods . . . . . . . . . . . . . . . . . . . .
6.3.1. The scenario analysis in GLOWA Jordan River . . .
6.3.2. Derivation of value functions for preference ranking .
6.3.3. Model validation . . . . . . . . . . . . . . . . . . .
6.4. Simulation experiment . . . . . . . . . . . . . . . . . . . .
6.5. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.6. Discussion and conclusions . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Jordan River
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
7.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2. Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2.1. Input data and model initialization . . . . . . . . . . . . . . . . .
7.2.2. Geographic distribution of vascular plant diversity . . . . . . . . .
7.2.3. Estimation of area potentials for agriculture and their spatial correlation to regions with high vascular plant diversity . . . . . . . . .
7.2.4. Assessment of land-use and land-cover change impacts on habitat
loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.2.5. Scenario drivers . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.3.1. Area potentials for agricultural activities . . . . . . . . . . . . . .
7.3.2. Scenario analysis . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
x
47
48
48
49
50
50
51
51
52
52
57
57
58
59
59
62
63
65
66
71
75
75
76
76
77
78
79
79
80
80
84
90
Table of contents
7.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8. Synthesis
8.1. Summary of findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1.1. Modeling the feedback between stocking density and biomass productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1.2. Quantifying the environmental impact of grazing in Jordan . . . . .
8.1.3. Future land-use and land-cover change scenarios for the Jordan
River region . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1.4. Assessment of future conflicts between agricultural land use and
biodiversity in Africa . . . . . . . . . . . . . . . . . . . . . . . . .
8.2. Outlook on further research . . . . . . . . . . . . . . . . . . . . . . . . .
9. Bibliography
95
95
96
96
97
98
99
101
A. Nonlinear correlation functions between stocking density and biomass
productivity
117
B. Input specification
119
C. Land-use and land-cover maps
123
xi
List of figures
2.1. Framework for understanding land-use/cover situations . . . . . . . . . . .
3.1. Overview of the spatial scale levels, interlinkages between the scale levels,
and state variables for the global-scale LandSHIFT version . . . . . . . . .
3.2. Overview of the spatial scale levels, interlinkages between the scale levels,
and state variables for the regional-scale LandSHIFT version . . . . . . . .
3.3. Structure of the integrated modeling system LandSHIFT . . . . . . . . . .
3.4. Generalized model structure of spatially explicit land-use change models . .
3.5. Examples for neighborhood definitions . . . . . . . . . . . . . . . . . . . .
3.6. Flowchart of land allocation for METRO . . . . . . . . . . . . . . . . . .
3.7. Flowchart of land allocation for AGRO . . . . . . . . . . . . . . . . . . .
3.8. Flowchart of land allocation for GRAZE . . . . . . . . . . . . . . . . . . .
4.1. Spatial distribution of mean annual temperature and precipitation in the
Jordan River region for the climate normal period 1961-1990 . . . . . . . .
4.2. Land-use and land-cover maps for the simulation runs with sustainable rangeland management, intensive rangeland management without productivity
reduction in case of overgrazing, and intensive rangeland management with
a productivity reduction of 10 % in case of overgrazing . . . . . . . . . . .
4.3. Area development of urban/built-up area and arable land for Israel, Jordan,
and the Palestinian National Authority . . . . . . . . . . . . . . . . . . .
4.4. Comparison of additional rangeland demand as compared to the year 2000
4.5. Unmet feed demand for the Palestinian National Authority . . . . . . . . .
4.6. Comparison of simulated rangeland area for the years 2000 and 2005 with
FAOSTAT data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.
5.2.
5.3.
5.4.
Study region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Land-use and land-cover maps for the years 2000 and 2050 . . . . . . . .
Development of simulated urban/built-up area, arable land, and rangeland
Spatial distribution of relative HANPP for the years 2000 and 2050 . . . .
7
12
13
15
16
25
26
29
30
36
40
41
42
43
43
50
53
54
55
6.1. Study region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.2. Workflow of the simulation experiment on future land-use and land-cover
change scenarios for the Jordan River region . . . . . . . . . . . . . . . . 63
xiii
List of figures
6.3. Relative operating characteristic curves for the three land-use activities METRO, AGRO, and GRAZE . . . . . . . . . . . . . . . . . . . . . . . . . .
6.4. Fraction of the land-use activities in the state areas given as percentage,
for the four GLOWA Jordan River scenarios . . . . . . . . . . . . . . . . .
6.5. Absolute area in 2000 and 2050 for the land-use activities METRO, AGRO,
and GRAZE, for the four GLOWA Jordan River scenarios . . . . . . . . . .
6.6. Aggregated land-use and land-cover change trends in 2050 for the four
GLOWA Jordan River scenarios . . . . . . . . . . . . . . . . . . . . . . .
6.7. Land-use and land-cover maps for the base year and for the year 2050, for
the four GLOWA Jordan River scenarios . . . . . . . . . . . . . . . . . . .
7.1. Suitability maps under baseline conditions for the agricultural activities crop
production under rain-fed conditions, crop production under irrigated conditions, and livestock grazing . . . . . . . . . . . . . . . . . . . . . . . .
7.2. Land-use and land-cover map for the base year . . . . . . . . . . . . . . .
7.3. Distribution of area potentials over the diversity zones, for crop production
under rain-fed conditions, crop production under irrigated conditions, and
livestock grazing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4. Distribution of the average suitability values and the corresponding standard
deviations for crop production under rain-fed conditions, crop production
under irrigated conditions, and livestock grazing over the diversity zones . .
7.5. Division of the area potentials, located in the different diversity zones, for
crop production under rain-fed conditions, crop production under irrigated
conditions, and livestock grazing into the four suitability categories . . . .
7.6. Distribution of area potentials, located on areas classified as rangeland
under baseline conditions, for crop production under rain-fed conditions,
crop production under irrigated conditions, and livestock grazing over the
diversity zones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.7. Distribution of the area demand calculated for the land-use activity METRO
over the diversity zones and share of these areas in the total area of the
different diversity zones . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.8. Distribution of the area demand calculated for the land-use activity AGRO
over the diversity zones and share of these areas in the total area of the
different diversity zones . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.9. Distribution of the area demand calculated for the land-use activity GRAZE
over the diversity zones, for baseline conditions as well as for the four
simulation runs in the year 2020 and share of these areas in the total area
of the different diversity zones . . . . . . . . . . . . . . . . . . . . . . . .
65
68
69
70
72
81
82
83
84
85
86
88
89
91
A.1. Nonlinear correlation functions between biomass productivity and stocking
density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
xiv
List of figures
C.1. Land-use and land-cover map for 2020 and map of changes in land use and
land cover as compared to baseline for simulation run one . . . . . . . . .
C.2. Land-use and land-cover map for 2020 and map of changes in land use and
land cover as compared to baseline for simulation run two . . . . . . . . .
C.3. Land-use and land-cover map for 2020 and map of changes in land use and
land cover as compared to baseline for simulation run three . . . . . . . .
C.4. Land-use and land-cover map for 2020 and map of changes in land use and
land cover as compared to baseline for simulation run four . . . . . . . . .
124
125
126
127
xv
List of tables
3.1. Land-use transition matrix . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2. Per capita area demand as a function of population density and development
status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3. Crop types implemented in the global-scale LandSHIFT version and their
attribution to the related FAO categories . . . . . . . . . . . . . . . . . .
3.4. Mean annual precipitation categories, applied in the regional version of
LandSHIFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.1. Share in land area of the major land-use categories for Israel, Jordan, and
the Palestinian National Authority . . . . . . . . . . . . . . . . . . . . . .
4.2. Land-use and land-cover types implemented in LandSHIFT.R and mapping
scheme for relating these to the IGBP Land Cover Legend . . . . . . . . .
4.3. Processed input to LandSHIFT.R for the Millennium Ecosystem Assessment
scenario “Order from Strength” . . . . . . . . . . . . . . . . . . . . . . .
4.4. Simulated rangeland extent for Israel, Jordan, and the Palestinian National
Authority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
27
28
31
35
37
39
42
5.1. Classification of land-use and land-cover types for the analysis with landscape pattern metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.1. Summary of quantified scenario assumptions . . . . . . . . . . . . . . . . 62
6.2. Composition and parameterization of suitability factors and land-use constraints for the land-use activities METRO, AGRO, and GRAZE . . . . . . 64
6.3. Urban/built-up area, area covered with fruits, vegetables, cereals/field crops,
and other crops, and rangeland area, as well as the changes in extent compared to the extent in 2000 for the four GLOWA Jordan River scenarios . . 67
7.1. Suitability factors and weights for the assessment of area potentials for
agricultural activities in Africa . . . . . . . . . . . . . . . . . . . . . . . . 78
7.2. Division of the area potentials under baseline conditions for crop cultivation
under rain-fed conditions, crop cultivation under irrigated conditions, and
livestock grazing into the four suitability categories . . . . . . . . . . . . . 81
7.3. Division of the area potentials, located on areas classified as rangeland under
baseline conditions, for crop cultivation under rain-fed conditions and crop
cultivation under irrigated conditions into the four suitability categories . . 83
xvii
List of tables
7.4. Area statistics for METRO, AGRO, GRAZE, and natural vegetation under
baseline conditions and for the four simulation runs . . . . . . . . . . . . . 87
B.1. Summary of data requirements for the simulation experiment on feedback
effects between stocking density and green biomass production . . . . . . . 120
B.2. Summary of data requirements for the simulation experiment on land-use
and land-cover change scenarios for the Jordan River region . . . . . . . . 121
xviii
1. Introduction
1.1. Background
Over the last three centuries, human activities have significantly altered the Earth’s terrestrial surface. Between 1700 and 1990, global cropland area increased from 265 million
hectare to 1 471 million hectare and rangeland area increased from 524 million hectare
to 3 451 million hectare (Klein Goldewijk, 2001). Today, approximately 37 % of the land
surface is under agricultural use, either as cropland (11 %) or as rangeland (26 %) (FAO,
2009). The “Green Revolution”, starting in the 1960’s, has furthermore led to a enormous
intensification of agriculture (Tilman, 1998) and, as a results, has increased the global per
capita food supply (FAO, 2009).
Besides the provision of food, the world’s ecosystems accomplish a large number of
functions essential to mankind (Daily, 1997; Millennium Ecosystem Assessment, 2003).
These include regulation functions (e.g. soil retention or pollination), habitat functions
(e.g. nursery function), production functions (e.g. production of medical resources), and
information functions (e.g. recreation) (de Groot et al., 2002).
Agricultural intensification and the expansion of agricultural area can induce changes in
structure and resilience of ecosystems and thereby reduce their capacity to provide goods
and services in the long run (Foley et al., 2005; Chapin III et al., 2000). Examples for
consequences of intensive agricultural management are degradation of water quality due to
elevated fertilizer input (e.g. Matson et al. (1997); Bennett et al. (2001)), loss of arable land
due to soil salinization induced by irrigation (Wood et al., 2000), or soil degradation through
intensive livestock grazing (Steinfeld et al., 1998). Moreover, expansion of agricultural area
causes loss and fragmentation of natural ecosystems and hence induces species extinction,
especially in regions with high species richness or endemism (Ceballos and Ehrlich, 2002;
Balmford and Long, 1994). In addition, agricultural intensification and the expansion of
agricultural area affect the energy and matter cycles of the earth system and, as a result,
the regional and global climate (Pielke Sr et al., 2002; Kalnay and Cai, 2003). In turn,
changing climate conditions influence species distribution (Walther et al., 2002) and are
likely to cause changes in crop yields and net primary productivity in many parts of the
world (Tubiello et al., 2007).
Over the next forty years, the world population is expected to increase from currently
6.8 billion to about 9 billion (United Nations, 2009a). In combination with an increasing
per capita consumption (Myers and Kent, 2003), this is likely to result in a doubling of the
world food demand by 2050 (Alexandratos, 1999; Bongaarts, 1996). Agricultural activities,
1
1. Introduction
essential to meet this food demand, will inevitably put additional pressure on natural
ecosystems as well as on existing agro-ecosystems. This effect is likely to be amplified by an
increasing cultivation of crops for the production of industrial raw materials and bioenergy
(Field et al., 2008). Hence, there is an urgent need for agricultural management strategies
that support the combination of agricultural activities with conservation of natural habitats
and resource-conserving agriculture, in order to maintain biodiversity and the provisioning
of ecosystem functions for future generations (Norris, 2008; DeFries and Eshleman, 2004).
Computer-based, mathematical simulation models of land-use change have emerged
as an essential tool for the analysis of the complex structures of land-use systems and
the interlinkages of the components of these systems (Verburg et al., 2004; Heistermann
et al., 2006; Schaldach and Priess, 2008). In order to be suited for the development
and evaluation of resource-conserving agricultural management strategies, these simulation
models of land-use change have to fulfill a set of formal requirements: Besides the ability
to assess the location of change, they have to be able to calculate changing levels of
intensification (Lambin et al., 2000). Furthermore, they have to operate on multiple spatial
scale levels, in order to bring together policy decision at the aggregated level (e.g. the
Common Agricultural Policy reform in Europe) with local impacts.
Scenarios, which are plausible descriptions of how the future may unfold (Alcamo, 2008),
have become a popular tools to estimate the impacts of human activities on future states
of the environment and the climate system. Important examples of scenario studies are
the Millennium Ecosystem Assessment (Carpenter et al., 2005), the fourth Global Environmental Outlook Assessment (Rothman et al., 2007), and the IPCC Assessment (IPCC,
2007). The application of land-use change models, in the context of scenarios studies can
support the development of strategies to minimize loss, fragmentation, and degradation of
ecosystems and hence help to identify ways to ensure the provision of food for a growing
and more demanding human population in the long run.
1.2. Objective
The overall objective of this thesis was to investigate management strategies for Africa
and the Jordan River region that allow a resource-conserving production of agricultural
commodities. The investigation was performed as a set of simulation experiments, focusing on different aspects of nature conservation and resource-conserving agriculture. The
simulation experiments were carried out as mid- to long-term (20 to 50 years) scenario studies applying the integrated modeling system LandSHIFT, which complies with the above
stated requirements (Schaldach and Koch, 2009).
The projected population increase for Africa is the highest worldwide (United Nations,
2009a). Furthermore, the African continent features large regions with high degrees of
biodiversity and endemic species (Orme et al., 2005). In order to investigate the reconcilability of food production to fulfill human demands and biodiversity conservation in Africa,
the area potentials for agricultural activities were calculated and a set of continental-scale
2
1.3. Structure of the thesis
biodiversity conservation strategies were tested. As a prerequisite for this simulation experiment, the global-scale LandSHIFT version had to be refined: One of the core processes
of LandSHIFT, the evaluation of local suitability for land-use activities, was expanded and
now considers the degree of biodiversity. The enhanced version is capable to gradually
constrain land-use activities in regions with high degrees of biodiversity.
The dryland ecosystems in the Jordan River region are highly vulnerable to desertification1 (Eswaran et al., 2001). One major cause of desertification is mismanagement
of livestock grazing (Steinfeld et al., 2006; Geist and Lambin, 2004). Permanently high
stocking densities may cause alterations in vegetation cover and composition (Gillson and
Hoffman, 2007), soil degradation (Ibáñez et al., 2007), and biodiversity reduction (Alhamad, 2006; Alados et al., 2004). According to Tsioras et al. (2006) one the most effective
approaches for fighting desertification is to avoid the removal of the vegetation cover by
adverse human activities such as overgrazing. In order to investigate the grazing systems
in the Jordan River region and to analyze strategies for reduction of land degradation
effects, a regional version of the integrated modeling system LandSHIFT had to be constructed. For this purpose, the global-scale LandSHIFT version was refined and adapted
to the biophysical and socio-economic conditions of the Jordan River region. The enhanced regional-scale LandSHIFT version, named LandSHIFT.R, now operates on a spatial
resolution of 1km, implements a representation of the feedback between stocking intensity
and biomass productivity, and provides the possibility to specify rangeland management
strategies that differ in their upper limit of stocking densities. LandSHIFT.R was then applied in three simulation experiments in order to test the new implemented functionalities,
develop methods to quantify the environmental impact of different grazing intensities, and
finally, investigate potential future development pathways for the region in the context of
an environmental scenario analysis.
1.3. Structure of the thesis
Chapter 2 provides an overview of the terminology and concepts that form the basis of
this thesis. This includes a definition of the most important terms, a description of causes
and consequences of changes in land-use systems, and an overview of common approaches
to model land-use and land-cover change.
Chapter 3 provides a comprehensive description of the integrated modeling system
LandSHIFT that aims at the simulation and analysis of spatially explicit land-use dynamics
and the resulting impact on the environment. The details of both LandSHIFT versions, the
global-scale LandSHIFT version and the regional-scale LandSHIFT version, are explained.
This chapter includes a description of the implementation of the feedback between stocking
intensity and biomass productivity, the different rangeland management strategies, and the
1
According to UNCCD (1994) desertification is “land degradation in arid, semi-arid and dry sub-humid
areas resulting from various factors, including climatic variations and human activities”.
3
1. Introduction
consideration of biodiversity in the calculation of the suitability for agricultural production
that were implemented into the model in the context of this thesis.
Chapter 4 describes a sensitivity analysis, which was conducted in order to test the
newly implemented functionalities that represent the feedback between stocking intensity
and biomass productivity as well as the different rangeland management strategies. Moreover, the impact of this feedback and of the rangeland management strategies on the
extent of rangeland was tested.
Chapter 5 presents a simulation experiment in which the enhanced regional-scale LandSHIFT version was used to calculate a “business-as-usual”-scenario with two different rangeland management strategies. Furthermore, it is described, how a set of landscape pattern
metrics and a biodiversity indicator can be combined in order to quantify the impact of different rangeland management strategies, and thus grazing intensities, on the fragmentation
of the landscape.
Chapter 6 describes the application of LandSHIFT.R in order to develop comprehensive
land-use and land-cover change scenarios for the Jordan River region in the context of an
environmental scenario analysis. The scenarios, including different rangeland management
strategies, help to investigate the impact of alterations in socio-economic conditions on
the landscape.
Chapter 7 shows a simulation experiment that applied the enhanced global-scale LandSHIFT version in order to quantify the potential agricultural area in Africa and the correlation of this area to spatial patterns of vascular plant diversity. It was analyzed to what extent
the inclusion of new conservation areas can help to reduce habitat loss. Moreover, the possible impact of future agricultural development on biodiversity and the effects of potential
biodiversity protection strategies on land-use and land-cover changes were investigated.
In addition to the thematic discussions provided in chapters 4 to 7, chapter 8 gives a
summary of the findings of these chapters. Furthermore, research needs emerging from the
integrated consideration of these simulation experiments are highlighted.
4
2. Land-use and land-cover
change: terminology and
concepts
The purpose of this chapter is to provide definitions of the most important terms and
a description of the underlying concepts of this thesis. This includes the definitions of
the terms land use, land cover, land-use change, land-cover conversion, and land-cover
modification. Additionally, a short summary of the causes and consequences of land-use
and land-cover change and an overview of the main approaches to modeling land-use and
land-cover change is given.
2.1. Definitions
It is important to clarify the terminology used in the context of this thesis. In the following,
the definitions of the terms land cover, land-cover conversion, land-cover modification, land use, and land-use change are given, and the interlinkages between them
are discussed.
Land cover. According to Lambin et al. (2000), the term land cover refers to the attributes of the Earth’s land surface and its immediate subsurface. These attributes
embrace biota, soil, topography, surface and groundwater, and man-made features
such as built-up structures.
Land-cover change. In general, changes in land cover can be divided into land-cover
conversions and land-cover modifications (Meyer and Turner II, 1994). Land-cover
conversion describes the displacement of one land-cover type by another, e.g., by
agricultural expansion or deforestation. In contrast, land-cover modifications induce
a change in the character but not in the type of land-cover (Lambin et al., 2003),
e.g., by agricultural intensification.
Land use. Land use, defined as “the purpose for which humans exploit the land cover”
(Lambin et al., 2000), involves two aspects: On the one hand the intent for the
manipulation of land cover and on the other hand the kind of manipulation (Lambin
and Geist, 2006). Examples for land uses are crop cultivation, livestock herding or
recreation.
5
2. Land-use and land-cover change: terminology and concepts
Land-use change. According to Meyer and Turner II (1994), the term land-use change
refers either to a shift from one land use to another or an increase in the land-use
intensity.
Meyer and Turner II (1994) see no clear attribution of one land-use type to one land-cover
type, even though a single land use may correspond to a single land cover (e.g. grazing to
unimproved grassland); one land-use type may involve multiple land-cover types and vice
versa. Changes in land use often result in changes in land cover.
2.2. Causes and consequences of land-use and
land-cover change
In the scientific literature, the relation between human land-use activities and the environment is often described as a coupled human-environment system with social as well as
ecological components (Mather, 2006). Accordingly, changes in land use and land cover
result from multiple interacting factors, including bio-physical drivers (e.g. changes in soil
fertility) and socio-economic drivers (e.g. economics and policy) (Fig. 2.1).
The causes of land-use and land-cover change can be divided into proximate (direct)
and underlying (indirect) causes (Geist and Lambin, 2002). The former, operating at the
local level, reveal how and why land cover and ecosystem processes are modified directly by
humans. Examples are agricultural expansion or the extension of infrastructure. The latter,
operating at regional to global level, represent the fundamental forces that give rise to
local actions. In general, underlying causes operate from a distance, often by changing one
or more proximate causes. Most important underlying causes of land-use and land-cover
change include natural variability, economic factors, technological factors, demographic
factors, institutional factors, cultural factors, and globalization (Lambin and Geist, 2007).
Land use profoundly affects the Earth’s ecosystems (Foley et al., 2005). The intensification of agriculture and the expansion of agricultural area have caused strong environmental
damage. Examples are the degradation of water quality due to fertilizer use (Matson et al.,
1997; Bennett et al., 2001; Wood et al., 2000), the loss of native habitats causing species
extinction and degradation of services of pollinators (Ceballos and Ehrlich, 2002; Balmford
and Long, 1994; Kremen et al., 2002), or declining groundwater tables and river discharge
through increased water withdrawals for agricultural activities, such as irrigation (Rosegrant
et al., 2002a). In addition, agricultural intensification and the expansion of agricultural area
affect the energy and matter cycles of the Earth system and, as a result, the regional and
global climate (Pielke Sr et al., 2002; Kalnay and Cai, 2003). A comprehensive overview
of global consequences of land use is given by Foley et al. (2005). Overviews of the consequences of agricultural intensification and agricultural expansion are provided by Matson
et al. (1997), Tilman (1999), and Tilman et al. (2002). DeFries and Eshleman (2004) and
Scanlon et al. (2007) summarize the impacts of land-use and land-cover change on water
resources and hydrological processes.
6
2.3. Main approaches to modeling land-use and land-cover change
Fig. 2.1.: Framework for understanding land-use/cover situations (Turner II et al., 1995).
2.3. Main approaches to modeling land-use and
land-cover change
According to Lambin et al. (2000), modeling of land-use and land-cover change should
aim to answer at least one of the following questions: Why, where, and when do land-use
and land-cover changes occur? Depending on the addressed research question, the most
suitable modeling approach has to be chosen.
Over the past years, many models addressing questions of land-use and land-cover change
have been developed. These models can be categorized according to the underlying mo-
7
2. Land-use and land-cover change: terminology and concepts
deling approach:
Cellular automata (CA) models are a common modeling approach for the spatial representation of land-use dynamics. According to Wolfram (1984), CA, defined as
discrete dynamical systems, contain a large number of identical basic components
that interact locally. These components, each with a limited set of possible values,
form a regular grid, which represents the modeling area. Starting from an initial system state, the values of the components evolve in discrete time steps on the basis of
local transition functions. The state change of a particular cell is determined by the
values of the neighboring components and its own value in the previous time step.
Cellular automata models typically operate at various spatial and temporal scales. In
land-use models, CAs often consider exogenous factors (White and Engelen, 1997).
Agent-based models describe the decision making of actors in the analyzed system. As
with CA models, the dynamics, simulated with an agent-based model exclusively
depends on the behavior of the single components, i.e. the agents. The autonomous agents, representing entities as, for instance, people or organizations, share
an environment by communicating and interacting. Unlike CA models, which focus
on landscape and transition, agent-based models emphasize human actions (Parker
et al., 2003).
Empirical/statistical models aim at identifying explicitly the causes of land-use and
land-cover change (Lambin et al., 2000). This type of model typically applies regression techniques using linear or logistic assumptions, in order to evaluate observed
land-use and land-cover changes (Schneider and Pontius Jr., 2001; Veldkamp and
Fresco, 1996). The evaluation can be performed for both, biophysical variables (e.g.
slope or NPP) and socio-economic variables (e.g. population growth). Since empirical/statistics models are only able to predict patterns represented in the calibration
data set, the transferability to other regions and the applicability to simulate land-use
intensity is limited (Lambin et al., 2000).
Stochastic models base upon transition probability models. They provide stochastic descriptions of processes moving in sequential steps through a set of states, whereas
the states are defined as the amount of land covered by differing land uses. The
transition probabilities are derived from statistical evaluation of observed transitions
(Lambin et al., 2000).
Optimization models originate from economics and base upon either the microeconomic or the macroeconomic approaches (Kaimowitz and Angelsen, 1998). Microeconomic approaches attempt to analyze resource use from an individual’s perspective,
considering economic factors such as prices or access to infrastructure. This might be
accomplished by agent-based models (Schreinemachers and Berger, 2006). The optimization procedure bases upon an objective function under consideration of boundary
8
2.3. Main approaches to modeling land-use and land-cover change
conditions, e.g., political constraints, often implemented using linear programming
(Hilferink and Rietveld, 1999). Macroeconomic models, operating on countries or
regions, are built on general or partial equilibrium sets of equations. These models
usually consider land resources as production factors. Spatial heterogeneity of land
is either ignored or included via yield functions.
Dynamic simulation models base on the assumption that spatial and temporal patterns of land-use and land-cover change are induced by the interaction of biophysical
and socio-economic processes. Hence, this type of model aims at reproducing these
processes and their interplay. Therefore, dependent on a priori understanding, complex ecosystems are described by a small number of differential equations (Lambin
et al., 2000). Dynamic simulation models emphasize the temporal dynamics of landuse systems and describe competition between different land uses, path dependencies,
and fixed land-use trajectories. Thus, they can be used to explore future changes in
land-use or to analyze land-use change trajectories (Verburg et al., 2006). These
models are in most cases not formulated in a spatially explicit manner.
Integrated modeling approaches combine elements of the different modeling techniques described above, in a way most appropriate to answer the respective research
question (Lambin et al., 2000). Examples are GEOMOD2 (Hall et al., 1995) and the
CLUE family (Veldkamp and Fresco, 1996), which combine statistical techniques
with CA models, or integrated assessment models that couple existing model from
different disciplines, e.g. IMAGE (Alcamo et al., 1998).
The diversity of modeling approaches is reflected by the large amount of publications
aiming at reviewing and classifying the different approaches to model land-use change.
Lambin (1997) and Kaimowitz and Angelsen (1998) review deforestation models, Miller
et al. (1999) and U.S. EPA (2000) provide overviews of integrated urban models, and
Parker et al. (2003) and Bousquet and Le Page (2004) provide reviews for multi-agent
modeling approaches. Lambin et al. (2000) provide an overview of models for agricultural
intensification, Schaldach and Priess (2008) review integrated models of the land system,
and Heistermann et al. (2006) give an overview of global-scale land-use models. Furthermore Briassoulis (2000) and Verburg et al. (2004) provide broad overviews of land-use
models.
9
3. The integrated modeling system
LandSHIFT
In this chapter, a detailed description of the integrated modeling system LandSHIFT is
given. The description includes the distinctive features of both, the global-scale and the
regional-scale LandSHIFT version. The headings and order of this chapter follow the Overview, Design concepts, and Details standard protocol for model descriptions, suggested by
Grimm et al. (2006).
3.1. Overview
3.1.1. Purpose
The integrated modeling system LandSHIFT1 (Alcamo and Schaldach, 2006; Schaldach
and Alcamo, 2006) is designed to simulate the spatial and temporal dynamics of largescale land-use systems. It is used to explore the impact of alterations in socio-economic and
biophysical conditions on spatial distribution and intensity of land use and the feedback of
land-use changes on the society and the environment at the regional to global scale. Besides
the exploration of possible future developments of land-use systems, the land-use model
LandSHIFT serves as a tool to formalize knowledge about and gain new insights into the
functioning of land-use systems. LandSHIFT is suitable to test hypotheses about processes
and interlinkages within these systems. Furthermore, the model can help to promote the
understanding of land-use systems by identifying key processes and their interlinkages and,
hence, reveal demands for further research activities.
LandSHIFT ‘s main field of application is the simulation of spatially explicit, mid- to longterm (20 to 50 years) future scenarios of land-use change. These scenarios explore possible
trends in land use and support the visualization of alternative land-use configurations.
The spatial modeling of land-use change supports the identification possible hot spots of
change and conflicts for land resources. Thus, spatially explicit land-use change scenarios
enable scientific support for evaluation and formulation of sustainable land-use planning
and policies. For this reason, they promote informed decision making.
At present, two LandSHIFT versions exist, which differ from each other in the spatial
scale of application. One version operates on a spatial resolution of 5 arc minutes (approx.
10 km at the equator) and has a global extent. Initial applications of this LandSHIFT
1
Land Simulation to Harmonize and Integrate Freshwater Availability and the Terrestrial Environment
11
3. The integrated modeling system LandSHIFT
Fig. 3.1.: Overview of the spatial scale levels, interlinkages between the scale levels, and
state variables for the global-scale LandSHIFT version.
version, in the following referred to as global-scale LandSHIFT version, include mid- to
long-term scenario analyses on the (sub-)continental scale for Africa (Schaldach et al.,
2006; Weiß et al., 2009) and India (Schaldach et al., in review). The other LandSHIFT
version operates on a spatial resolution of 1 km and covers the Middle East (Koch et al.,
2008). This LandSHIFT version, in the following referred to as regional-scale LandSHIFT
version or LandSHIFT.R, was developed in the context of the GLOWA Jordan River project
(GLOWA JR), funded by the German Federal Ministry of Education and Research (BMBF).
Within the scope of GLOWA JR‘s scenario analysis, LandSHIFT.R serves for the simulation
of land-use change scenarios up to 2050.
3.1.2. State variables and scales
The representation of land-use systems in LandSHIFT is organized on interacting spatial
scale levels. On these scale levels, the state variables of the modeled land-use systems are
defined. The interlinkage between the scale levels is realized by an exchange of the state
variables. An overview of the spatial scale levels, the interlinkages between the levels and
the state variables for the two different LandSHIFT versions is given in Fig. 3.1 and 3.2.
12
3.1. Overview
Fig. 3.2.: Overview of the spatial scale levels, interlinkages between the scale levels, and
state variables for the regional-scale LandSHIFT version.
Macro level. The logical as well as the spatial configuration of the macro level is based on
countries. The state of a country is defined by the variables total human population
number, crop production in metric tonnes, total livestock number (ruminants), yield
change due to technological change for the modeled crop types, and fraction of crop
area under irrigation for the modeled crop types. All state variables specified on the
macro level, constitute driving forces of change in land-use systems.
Intermediate level. The intermediate level is represented by a regular grid with a spatial
resolution of 30 arc minutes; this corresponds to approximately 50 km at the equator.
The state variables, defined on this level, differ between the two LandSHIFT versions.
For the global-scale LandSHIFT version, the state of an intermediate-level grid cell
is defined by the potential yields of the modeled crop types under rain-fed as well as
under irrigated conditions and by the net primary productivity (NPP) of rangeland
and natural vegetation. For the regional-scale LandSHIFT version, the state of an
intermediate-scale grid cell is defined by the potential wheat yield under rain-fed and
irrigated conditions.
Ecoregion level. For Israel, the spatial-scale hierarchy of the regional-scale LandSHIFT
version was extended by a level based on so called ecoregions (Kan et al., 2007).
13
3. The integrated modeling system LandSHIFT
The incorporation of the additional scale level was essential to include information
on crop production with a spatial resolution below country level. Consequently, the
state of an ecoregion is defined by production in metric tonnes for the modeled crop
types.
Micro level. The geographic extent of each country is specified by the micro level, a
regular grid with a uniform cell size. Land-use and land-cover change are calculated
on the micro level. The global-scale and the regional-scale LandSHIFT versions differ
with respect to the cell size; the global-scale version features a cell size of five
arc minutes, the regional-scale version has a cell size of one kilometer. The state
of a micro-level grid cell is defined by dominant land-use/land-cover type, human
population density, stocking density and allocated crop production of the modeled
crop types. Furthermore, a set of quasi-static landscape characteristics and land-use
constraints are defined on the micro level. The landscape characteristics comprise
terrain slope, road infrastructure and river-network density. National and international
conservation areas are incorporated as land-use constraints. The global-scale LandSHIFT version features the additional land-use constraints Tsetse fly related disease
risk for livestock and zones with high vascular plant diversity. The regional-scale
LandSHIFT version has an additional state variable, describing the relative human
appropriation of net primary production (relative HANPP).
Both LandSHIFT versions currently use a 5-year time step. Consequently, simulations run
for 5 to 11 time steps (simulation period of 20 to 50 years). The length of the simulation
period depends on the research question the respective simulation experiment is supposed
to answer.
3.1.3. Process overview and scheduling
The processes implemented in LandSHIFT are organized in three modules, which operate on the different spatial scale levels by modifying the scale-specific state variables.
The Biophysic module, which describes the environmental sub-system, comprises process representations for the calculation of crop yields and NPP of rangeland and natural
vegetation (Fig. 3.3). These process representations operate on the macro level and, in
case of the regional-scale LandSHIFT version, additionally on the micro level. The Socioeconomy module is part of the representation of the human sub-system. This module
includes algorithms for the calculation of human population growth, agricultural production and trade (crop production and livestock number), as well as changes in crop yield
due to technological change. The processes of the Socio-economy module are formulated
on the macro level and, in case of the regional-scale LandSHIFT version, additionally on
the ecoregion level. In each time step, first the modules Socio-economy and Biophysic are
executed and the corresponding state variables are updated. Subsequently, information delivered by these modules is used by the Land-use change module (LUC-module), which
14
3.1. Overview
Fig. 3.3.: Structure of the integrated modeling system LandSHIFT, adapted from Schaldach and Koch (2009).
is also part of the representation of the human sub-system, to simulate land-use change.
The processes of the LUC-module are formulated on the micro level. The bidirectional
information exchange between the three modules is realized by state variables.
Based on the information provided by the Biophysic and the Socio-economy module, the
LUC-module accomplishes the calculation of quantity and location of land-use change. The
present realization of the LUC-module implements the land-use activities housing and infrastructure, crop production, and livestock grazing. The processes representing the different
land-use activities are organized in sub-modules: METRO for housing and infrastructure,
AGRO for crop production, and GRAZE for grazing. The competition between the landuse activities for land resources is addressed by a ranking of the activities according to their
relative economic importance. The ranking defines the sequence of execution, starting with
the most important land-use activity: METRO → AGRO → GRAZE. In one simulation
step, cells occupied by a superordinate land-use activity are not available for a subordinate
land-use activity.
In every simulation step, each sub-module executes the functional parts demand processing, preference ranking and land allocation. This corresponds to the generalized
structure of spatially explicit land-use change models (Fig. 3.4) as presented by Verburg
et al. (2006). First, within the demand processing part, driving forces of land-use change
are converted to macro-level and ecoregion-level demands for services and agricultural commodities. Second, within the preference-ranking part, the micro-level grid cells are valued
in terms of their suitability for the respective land-use activity, resulting in a suitability
15
3. The integrated modeling system LandSHIFT
Fig. 3.4.: Generalized model structure of spatially explicit land-use change models (Verburg
et al., 2006).
map. The grid cells are then ranked according to their suitability. Third, within the landallocation part, each land-use activity manipulates the micro-level state variables land-use
type, population density or livestock density in order to meet the demand for the commodity or service under consideration. The range and magnitude of change is constrained on
the one hand by the demand for the respective commodity or service, and on the other
hand by the supply, i.e. the productivity on the particular micro-level grid cell. As a result,
the land-use type, population density or livestock density of as many cells as required to
meet the demand has to be changed.
The output of LandSHIFT, i.e. the documentation of the simulation of land-use and
land-cover changes, comprises time series of micro-level maps displaying the dominant
land-use and land-cover type, human population density, crop production quantity and
stocking density. Moreover, the output includes a set of indicators, aggregated on the
macro-level (e.g. area statistics). The output of the regional-scale LandSHIFT version
additionally contains maps on the relative human appropriation of net primary production,
as defined by Haberl et al. (2007b).
3.2. Design concepts
The integrated modeling system LandSHIFT combines a set of different design concepts.
The choice of concepts is determined by the purpose of LandSHIFT, described in section
3.1.1. With regard to design concepts, LandSHIFT can be characterized as:
• Dynamic. According to Lambin et al. (2000), one of the questions the modeling
of land-use changes should address, is the point in time and the rate of changes.
16
3.2. Design concepts
A prerequisite for the representation of the complex temporal dynamics of land-use
systems is dynamic modeling (Verburg et al., 2006). LandSHIFT applies a dynamic
modeling approach, i.e. it subdivides the simulation period into several time steps.
Hence, LandSHIFT fulfills the basic requirement for the simulation of land-use change
trajectories, feedbacks and path dependencies in the evolution of land-use systems
and thus in principal can be applied in order to analyze when and with which rate
land-use changes occur.
• Integrated. Integrated modeling systems have to include information from more
than one discipline, organize this information in a modularized program structure,
and link scientific findings with policy analysis (Alcamo, 2002). LandSHIFT is an
integrated modeling system. It provides a framework for the combination of biophysical and socio-economic information with geographical information on land use
and land cover, and integrates this information, generated with different modeling
approaches, in form of a modularized program structure. This structure facilitates
a flexible exchange of functional modules and the easy extension with new modules
from the rapidly advancing field of land-use research, and hence meets the challenge
of the semantical and technical integration of different modeling approaches within
a common conceptual and software framework. The global-scale and the regionalscale LandSHIFT version are applied in the context of scenario analyses with political
relevance (e.g. GEO4 or GLOWA JR).
• Process based. LandSHIFT applies a process-based modeling approach. The LandSHIFT modeling system is an artificial representation of a coupled human-environment system (see description of land-use systems in chapter 2). The design of LandSHIFT enables the exploration of the interlinkages between these two subsystems.
LandSHIFT depicts the key processes that lead to changes in these systems. As pivotal process, LandSHIFT implements human-decision making regarding the quantity,
location and intensity of land-use activities. Hence, pattern in land-use and landcover change are predicted from process, also referred to as a top-down modeling
(Verburg et al., 2006). The application of the process-based modeling approach, makes LandSHIFT suitable for the simulation and analysis of the functioning of land-use
systems. Furthermore, it allows the analysis of trajectories and intermediate states
of land-use and land-cover change (Verburg et al., 2006), e.g. via scenarios.
• Spatially explicit. Spatially explicit land-use models simulate changes in land-use
for individual spatial entities (Verburg et al., 2006). In case of LandSHIFT, which
operates on multiple hierarchically organized spatial scale levels, these entities are
cells of a regular grid - the micro-level grid cells - and hence, LandSHIFT can be specified as pixel based model. Changes in land-use are modeled for these cells. Spatially
explicit models are able to simulate the location and spatial variability of land-use
and land-cover changes, and as a result enable the analysis of relations between social and biophysical environments, and variations in location and quantity of land-use.
17
3. The integrated modeling system LandSHIFT
Most spatially explicit models depend on the same general model structure (Fig.
3.4).
One result of LandSHIFT, and of spatially explicit land-use change models in general,
are maps displaying changes in land-use patterns. These maps help to reveal possible
hot-spots of land-use change and hence allow the identification of priority areas
for further investigations or focus areas for alternative management strategies, e.g.
in form of conservation areas. Furthermore, the spatially explicit simulation results
serve for the analysis of the competition for land resources (e.g. bioenergy-crop
production and food-crop production) or ecosystem services. The maps enable the
visualization of the land-use changes resulting from different configurations of model
input and parameterization, representing different political or societal decisions and
developments. This makes LandSHIFT a valuable tool in order to provide scientific
support for stakeholders involved in land-use decision making.
3.3. Details
3.3.1. Initialization
The basic land-use and land-cover map used by the global-scale LandSHIFT version was
generated with an allocation procedure that combines land-cover information derived by
remote sensing with census data on national and sub-national levels (Heistermann, 2006).
The map includes information on the spatial distribution of 17 major crop types as well as
information on the distribution of rangeland. In contrast, information on the spatial distribution of rangeland and the division of cropland into the different internal crop categories
is not included in the basic land-use and land-cover map employed by the regional-scale
LandSHIFT version.
In order to synchronize the information on population density and the land-use and landcover map information on urban areas, both LandSHIFT versions initially read their basic
land-use and land-cover maps from the data base. The basic land-use and land-cover map
is then combined with the micro-level information on population densities. The land-use
type of the micro-level grid cells, at which the population density exceeds the upper limit
for population density in rural regions and at which the land-use type is not assigned as
“urban”, is changed to “urban”. This initialization technique is applied for the land-use
activity housing and infrastructure in both LandSHIFT versions.
Since the processing mode for the basic land-use and land-cover map differs for the
two LandSHIFT versions (see above), the LandSHIFT versions apply different initialization
techniques for the land-use activities crop production and grazing. In order to initialize the
land-use activity crop production, the global-scale LandSHIFT version compares for each
internal crop type c the macro-level production derived from census data pcensc with the
sum of the local production on cells with that crop type in the initial land-use and landcover map as calculated by LandSHIFT pcalcc and calculates a crop-specific management
18
3.3. Details
parameter basec , defined as
basec = pcensc /pcalcc .
(3.1)
The purpose of this crop-specific management parameter is to account for uncertainties
caused by agricultural management strategies, such as multi cropping, which influence the
amount of crop production, but are not explicitly considered in LandSHIFT. The management parameters for the different crops are calculated for the base year of a simulation run
and are used in the following simulation steps for the correction of the crop productions.
For rangeland, a corresponding ratio is assessed in the base year and applied in order to
parameterize the share of grazing in fodder composition.
In order to initialize the land-use activity crop production, the regional-scale LandSHIFT
version first of all distributes the demands for the internal crop categories to the best suited
micro-level grid cells. The land-use type of grid cells, which are attributed as “cropland”
in the initial land-use and land-cover map but not assigned to one of the crop categories,
is converted to “other crops”. For the purpose of initializing the land-use activity grazing,
the regional-scale LandSHIFT version distributes either a given demand for biomass or an
area demand to the best suited micro level grid cells. The further proceeding for both landuse activities, corresponds to the one described for the global-scale LandSHIFT version. A
distinctive feature of the regional-scale LandSHIFT version is that the stocking densities
on the micro-level grid cells, can be restricted by the stocking capacities. The respective
modes of initialization, applied for the different simulation experiments, is described in the
corresponding chapters separately (see chapters 4 to 7).
3.3.2. Input
In general, the LandSHIFT input comprises of information on total human population,
crop production, yield change due to technological change, livestock numbers as well as
socio-economic information, e.g. on environmental policy or regional planning. Besides
this input specified on the macro level, LandSHIFT requires data on landscape and landuse characteristics on the micro level. A detailed description of the input is given for
each simulation experiment separately in the respective chapters and in the appendix (see
chapters 4 to 7 and appedix B).
3.3.3. Sub models
The processes implemented in LandSHIFT are organized in the three modules Biophysic
module, Socio-economy module and Land-use-change module. The details of these
modules are described below.
19
3. The integrated modeling system LandSHIFT
The Biophysic module
In each simulation step, the Biophysic module updates the state variables crop yields and
NPP of rangeland and natural vegetation, and provides these to the LUC-module. The
calculation of crop yields is based on the output of dynamic, process-based, terrestrial
ecosystem models. In the context of this thesis, the global-scale LandSHIFT version is
based on simulation results of the LPJmL model (Bondeau et al., 2007; Sitch et al., 2003)
and the regional-scale LandSHIFT version is based on simulation results of a gridded
version of the DAYCENT model (Parton et al., 1998; Stehfest et al., 2007). Both models
provide global maps of potential yields for the major crop types under rain-fed and irrigated
conditions, generated on the intermediate grid for current climate conditions. The yield
of a crop type on a grid cell is the yield under current climate conditions, corrected for
the effects of technological change. The yields under rain-fed and irrigated conditions are
weighted with a crop-specific fraction of irrigated area specified for each country separately.
The fraction of irrigated area is assumed to remain static for the simulation period.
The calculation of the state variable NPP of rangeland and natural vegetation for the
regional-scale LandSHIFT version is based on the output of WADISCAPE (Köchy, 2007;
Köchy et al., 2008). The global-scale LandSHIFT version employs the LPJmL model in
order to derive global maps of NPP of rangeland and natural vegetation. In contrast to
crop yield calculations, no impact of technological change is taken into account.
Although not applied in the context of this thesis, LandSHIFT is capable to include
information of crop yields and NPP of rangeland and natural vegetation under changing
climate conditions. This is realized by a correction for climate change based on a linear
interpolation between the productivities under current climate conditions and the respective
climate scenario.
DAYCENT The regional-scale LandSHIFT version is based on wheat yields under rainfed and irrigated conditions for the climate normal period 1961-1990, calculated with a
gridded version of DAYCENT (Parton et al., 1998; Stehfest et al., 2007). DAYCENT,
the daily version of the terrestrial ecosystem model CENTURY, simulates the dynamics
of carbon, nitrogen, phosphorus, and sulfur cycles in natural and agricultural systems. It
incorporates a detailed representation of plant growth, soil water fluxes and nutrient dynamics. Driving forces are, amongst others, precipitation, temperatures and agricultural
management. The calculation of potential crop yields is carried out as function of insolation, biomass, temperature and a crop-specific energy-biomass conversion factor. Potential
yields are limited by water and nitrogen availability. The information on crop yields is
available in grid format with a spatial resolution of 30 arc min. This information is geographically mapped to the micro level of the regional-scale LandSHIFT version without
further downscaling or correction (Fig. 3.2).
LPJmL The LPJmL model (Lund-Potsdam-Jena managed Land) provides the globalscale LandSHIFT version with global maps of crop yields and NPP of rangeland and
20
3.3. Details
natural vegetation for the climate normal period 1961-1990. LPJmL, based on the LPJ
dynamic global vegetation model (Sitch et al., 2003), includes a detailed representation
of plant growth, soil-water fluxes and carbon cycles. The distribution and growth of plant
functional types and crop functional types is based on bioclimatic limits. LPJmL simulates
the productivity of 10 natural plant functional types and the productivity and yield 13 crop
functional types (11 arable crops and 2 managed grass types). Crop yields are calculated
for rain-fed as well as irrigated conditions. The information generated with LPJmL in a
spatial resolution of 30 arc minutes, is geographically mapped to the micro level of the
global-scale LandSHIFT version without further downscaling or correction (Fig. 3.1).
WADISCAPE The WADISCAPE model (Köchy, 2007; Köchy et al., 2008) provides
the regional-scale LandSHIFT version with information on stocking capacities2 as well as
information on the relationship between stocking density with small ruminants (goats and
sheep) and productivity of green biomass3 under current climate conditions. WADISCAPE
simulates the growth and dispersal of herbaceous plants and dwarf shrubs in artificial, fractal
wadi landscapes (wadiscapes) of 1.5 km × 1.5 km. Main exogenous driver of vegetation
dynamics in WADISCAPE is water availability, which is derived from precipitation under
consideration of slope angle, aspect and topographic position. The simulation of vegetation
dynamics is based on validated small-scale models of annuals (Köchy, 2006; Köchy et al.,
2008) and dwarf-shrubs (Malkinson and Jeltsch, 2007). WADISCAPE simulations were
conducted for five climatic regions (arid to mesic Mediterranean climates of the Middle
East) and, in factorial combination, five varying slope categories (0◦ to 30◦ ). In order to
determine the stocking capacity of the vegetation, these simulations were repeated for
stocking densities ranging from 0 to 10 animals per hectare.
The Socio-economy module
The Socio-economy module operates on the macro level and, in case of the regionalscale LandSHIFT version, additionally on the ecoregion level. The module accounts for
the organization and processing of the state variables human population number, crop
production, livestock number, and changes in crop yield due to technological change. For
historical periods, the information on the state variables is derived from census data. For
future periods, within the scope of scenario analyses, this information is provided by a set
of different approaches and models. An update of the state variables is carried out by the
Socio-economy module in each simulation step.
In the context of this thesis, the global-scale LandSHIFT version is based on future
human population numbers provided by modeling system International Futures (Hughes
2
The stocking capacity of a habitat is defined as the number of small ruminants per hectare for which
the vegetation provides sufficient food in 9 of 10 years in year-round grazing.
3
Green biomass is referred to as the aboveground biomass of herbaceous plants and leaf mass of dwarf
shrubs.
21
3. The integrated modeling system LandSHIFT
and Hillebrand, 2006) and on possible future developments of crop production, yield changes due to technological change, and livestock numbers calculated with IMPACT model
(Rosegrant et al., 2002b, 2005b). The regional-scale LandSHIFT version uses information
on human population number generated with the SAS (Story And Simulation) approach
(Alcamo, 2008), and information on yield increase due to technological change calculated
with IMPACT. Depending on the simulation experiment, information on crop production
was provided by IMPACT or by the VALUE model (Kan et al., 2007), and information on
livestock numbers was generated with SAS or IMPACT (see chapters 4 to 7).
IFs The global-scale LandSHIFT version uses human population numbers, calculated
with the large-scale, integrated modeling system International Futures (IFs). The modeling
system applies a dynamic, equilibrium seeking modeling approach and thus enables longterm scenario simulation. IFs represents demographic, economic, energy, agricultural, sociopolitical, and environmental subsystems for 183 countries, interacting on the global scale.
IMPACT Both, the regional-scale and the global-scale LandSHIFT versions use information on crop production, livestock number, and yield increase due to technological change,
provided by the International Model for Policy Analysis of Agricultural Commodities and
Trade (IMPACT). IMPACT, a representation of a competitive global market for agricultural commodities, is designed for the analysis of current conditions and possible future
developments in food demand, supply, trade, income and population. The model covers 32
commodities and 36 countries/regions, linked through trade, that account for almost all of
the world food production and consumption. IMPACT is based on a system of supply and
demand elasticities implemented via linear and nonlinear equations. It incorporates demand
as a function of prices, income and population growth, and changes in crop production.
The changes in crop production are determined on the basis of crop prices and productivity
growth rates.
SAS The SAS (Story And Simulation) approach to scenario development combines both,
quantitative and qualitative aspects of scenarios (Alcamo, 2001, 2008). The combination of
theses two aspects makes the resulting scenarios on the one hand generally understandable
and on the other hand suitable for planning purposes. Distinctive features of SAS are the
iterative structure and the intensive participation of experts and stakeholders. A detailed
description of SAS and its application in the context of a scenario analysis for the Jordan
River region is given in chapter 6. In the simulation experiment, delineated in chapter 6,
SAS was applied to generate information on human population number as input for the
regional-scale LandSHIFT version. Furthermore, qualitative scenario information was used
to select further drivers of land-use and land-cover change for the regional-scale LandSHIFT
version.
22
3.3. Details
VALUE In the context of this thesis, VALUE simulation results on ecoregion-specific crop
productions were used in the simulation experiment on land-use and land-cover scenarios for
the Jordan River region (chapter 6). The economic, nonlinear mathematical programming
model VALUE is designed for the evaluation of climate change impacts on agriculture at
the regional scale. It provides the regional-scale LandSHIFT version with information on
crop production for the different ecoregions in Israel (Fig. 3.2). VALUE is based on the
production-function approach and thus applies response functions of yield to describe the
impact of climate factors on agricultural productivity. The only climatic factor taken into
account in VALUE is precipitation. Besides, VALUE includes the effect of irrigation and
salinity of applied water via a yield-water-salinity response function. This allows to consider
not only for the effect of the quantity of applied water, but also for its quality. The model
reproduces the optimal behavior of farmers at the regional scale by maximizing net returns
from agricultural production, restricted by land and irrigation water availability.
The Land-use change module
The LUC-module is the central component of LandSHIFT and accomplishes the simulation
of the location and quantity of land-use changes. This is realized by a regionalization of
the macro-level (and ecoregion-level) demands for area intensive services, such as housing
and infrastructure, and agricultural commodities to the micro level. The basic principle is
to allocate the demands to the most suitable micro-level grid cells by changing the landuse type, population density or livestock density of as many cells as required to meet the
demand. Each commodity and service is linked to a land-use type, i.e., is supplied on grid
cells with the corresponding land-use type. The current version of the LUC-module implements the sub-modules METRO (housing and infrastructure), AGRO (crop production),
and GRAZE (grazing). In each simulation step, the sub-modules are executed subsequently and each of this sub-modules executes the three functional parts demand processing,
preference ranking and land allocation. In the following, the general operating mode of the
functional parts is described, followed by a description of the individual characteristics of
these for the different land-use activities.
Demand processing The functional part demand processing is responsible for the transformation of the drivers of land-use change to macro-level (and ecoregion-level) demands for the implemented services and commodities.
Preference ranking The functional part preference ranking operates on the micro level
and serves for the identification and ranking of the most suitable grid cells for the
different land-use activities and the corresponding land-use types. A method from
the field of Multi-Criteria Analysis (Cromley and Hanink, 1999; Eastman et al., 1995)
is applied in order to calculate the suitability values of a grid cell for the different
23
3. The integrated modeling system LandSHIFT
land-use types. The preference value Ψk is calculated as
Ψk =
n
X
wi fi (pi,k ) ×
gj (cj,k )
(3.2)
j=1
i=1
|
m
Y
{z
suitability
}
|
{z
constraint
}
with i wi = 1 and fi (pi,k ), gj (cj,k ) ∈ [0, 1]. The equation consists of two terms.
The first term of the equation is the sum of the weighted suitability factors pi ,
contributing to the suitability of a grid cell k for a particular land-use type. The
weight wi determines the importance of a single suitability factor in the analysis.
The second term of the equation, appended by multiplication, is the product of
the land-use constraints cj . These constraints reflect important aspects of humandecision making, e.g. the restriction of land use in conservation areas.
P
The suitability factors, their weights and the land-use constraints are specified on
the macro level and implemented as time-dependent variables. This enables the
representation of changing political and environmental boundary conditions. The
suitability factors and land-use constraints are standardized by the value functions
fi and gj , respectively, to a co-domain from 0 to 1 (Geneletti, 2004). The value
functions can be defined as positive or negative relationships and are scaled by the
range of the respective factor within a country, in order to account for the countryspecific spatial heterogeneity of the suitability factors and land-use constraints.
Suitability factors and constraints can be state variables, landscape characteristics,
zoning regulations or spatial neighborhood characteristics (White and Engelen, 1997).
The neighborhood of the micro-level grid cells is analyzed in each simulation step in
order to generate statistical or boolean information about the land-use/land-cover
type of the adjacent cells. The neighborhood of a cell can be defined by type and
order, e.g. a von Neumann or a Moore neighborhood (Fig. 3.5). Additionally, a (geographic) search radius can be specified. In general, the set of relevant suitability
factors and land-use constraints, the types of value functions and the factor weights
can be derived by either data driven procedures (e.g. geostatistical analyses) or by
expert knowledge acquisition. The latter can be formalized by tools like the Analytical
Hierarchy Process (Saaty, 1996).
The land use is restricted in conservation areas. The global-scale LandSHIFT version
applies also a land-use constraint in regions featuring high vascular plant diversity.
One special constraint is the transition between the different land-use types implemented in LandSHIFT: not all land-use types can be converted into each other. This
is a common approach in the field of land-use modeling, see e.g. Foley et al. (2005)
and Verburg et al. (2002). A summary of possible conversions is given in Tab. 3.1.
Land allocation The functional part land allocation assigns the macro-level (and ecoregion-level) demands for services and agricultural commodities to the micro-level
24
3.3. Details
Fig. 3.5.: Examples for neighborhood definitions. Von Neumann neighborhood (left), Moore neighborhood with radius r = 1 (center), and Moore neighborhood with radius
r = 2 (right).
Tab. 3.1.: Land-use transition matrix. Possible conversions are indicated by +, impossible
conversions are indicated by -.
From / To
Urban land
Cropland
Rangeland
Set aside
Natural vegetation
Urban land
+
+
+
+
+
Cropland
+
+
+
+
Rangeland
+
+
+
Set aside
+
+
-
Natural vegetation
+
grid cells with the highest suitability for the associated land-use type. Each land-use
activity implements its own allocation strategy, described below.
METRO The sub-module METRO simulates the spatial and temporal dynamics of area
for housing and infrastructure. Changes in quantity and location of this area are induced
by alterations in the total human population number, specified on the macro-level.
The preference ranking for METRO considers as suitability factors the two landscape
characteristics terrain slope and available road infrastructure (Solecki and Oliveri, 2004).
The value functions define a negative relationship for terrain slope (i.e. the higher the
terrain slope on a grid cell, the less suitable the grid cell) and a positive relationship for
the available road infrastructure.
The land allocation for METRO distinguishes between municipal and rural regions (Fig.
3.6). Depending on the region, a different kind of growth process is applied. Therefore, the
micro-level grid cells are grouped into these two categories (Antrop, 2000). A municipal cell
is defined as a cell, that features land-use type “urban” or has at least one grid cell with the
land-use type “urban” in its direct neighborhood. All other cells are defined as rural cells.
In municipal regions, the growth of urban areas is implemented as urban encroachment
process (Wu, 1999), i.e., new area for housing and infrastructure is located at the edges
of existing urban area (Solecki and Oliveri, 2004). In order to allocate additional human
population, a three step procedure is applied. First, a parameter defines the fractions
of the additional population which are assigned to municipal regions and rural regions,
25
3. The integrated modeling system LandSHIFT
Fig. 3.6.: Flowchart of the functional part land allocation for METRO.
respectively. Second, depending on the grid cell’s actual population density and suitability
value, an additional population number is allocated. Cells with the land-use type “urban”
feature an upper threshold for population density, beyond that no additional population
is assigned to that particular cell. Third, based on the re-calculated population densities,
land-use conversions are calculated. In case, the population density in rural regions exceeds
a pre-defined threshold value, the land-use type of the respective grid cells is changed
to “urban”. In rural regions, each cell has a fraction that is occupied by housing and
26
3.3. Details
Tab. 3.2.: Per capita area demand as a function of population density and development
status, adapted from Erb et al. (2007).
Industrialized
countries
Developing
countries
Sparsely populated
(< 50 cap/km2 )
[m2 /cap]
Medium population density
(50–250 cap/km2 )
[m2 /cap]
Densely populated
(> 250 cap/km2 )
[m2 /cap]
1 000
500
300
150
75
75
infrastructure based on the local human population density. The new area for housing
and infrastructure is calculated based on the population density and the per capita area
demand, as given in Tab. 3.2.
AGRO The sub-module AGRO accomplishes the calculation of the quantity and location
of rain-fed and irrigated cropland. The global-scale LandSHIFT version distinguishes between 12 different crop types, focusing on annual crops. Table 3.3 summarizes these internal
crop types, and lists their relation to the FAO crop categories. The regional-scale LandSHIFT version distinguishes between three crop categories, described in detail in chapter
6. In order to process the crop demands, the crop production is corrected by a crop-specific
management factor. These management factors are calculated in the initialization of the
simulations, as the ratio of the local crop production, which is the sum of the crop production on all cells covered with the associated crop type, and the crop production specified
on the macro level. The purpose of this correction is to match the macro-level production
and the local production at the start of the simulation and through this account for uncertainties due to agricultural management, not explicitly considered in LandSHIFT (e.g.
multi-cropping), or discrepancies due to the aggregation of crop types.
Preference ranking is performed for each crop type separately. In order to calculate the
suitability of a micro-level grid cell for the respective crop type, the suitability factors crop
yield, terrain slope, proximity to residential area, and proximity to existing cropland are
taken into account. Crop yields are considered as proxy for soil fertility, assuming that a
high crop yield represents a high preference for a particular crop type. For terrain slope,
a negative relationship is defined: the higher the slope value, the higher the reduction of
usability due to limitations in workability and accessibility, and additionally a higher risk
a fertility losses due to topsoil erosion (Fischer et al., 2002). For proximity to residential
area, a positive relationship is defined, with the proximity representing accessibility of
infrastructure, market access and local demand for crops (von Thünen, 1966). The value
function for proximity to existing cropland also defines a positive relationship.
The basic principle of the land allocation in AGRO is to formulate a “compromise
solution”-problem for the determination of a quasi-optimum crop allocation in order to
deal with the competition between the different crop types for the resource land. This is
27
3. The integrated modeling system LandSHIFT
Tab. 3.3.: Crop types implemented in the global-scale LandSHIFT version and their attribution to the related FAO categories.
Internal crop type
Wheat
Temperate cereals
Rice
Maize
Tropical cereals
Pulses
Tropical roots and tubers
Temperate roots and tubers
Annual oil crops
Soybeans
Sugarcane
Other
Related FAO categories
Wheat
Barley, oats, rye, buckwheat
Paddy rice
Maize
Millet, sorghum
Dry beans, dry peas, chick peas, lentils
Cassava, sweet potatoes, yams
Potatoes
Groundnuts, rape, sesame, sunflower, etc.
Soybeans
Sugarcane
Cotton, fruits, vegetables, coffee,
cocoa, tea, tobacco, oil palms, coconut,
olives, etc.
implemented via of a modified version of the Multi Objective Land Allocation (MOLA)
heuristic (Eastman et al., 1995). The original version resolves emerging conflicts by a pair
wise comparison, cells claimed by more than one crop type are allocated to the crop type
with the higher suitability. For the application in LandSHIFT, the heuristic was modified
in two ways. First, instead of a given area, the modified version allocates crop demands.
Second, pattern stability is considered in the conflict resolution step, i.e., if no change in
crop demands occur, the land-use and land-cover pattern remains constant, even if better
suited grid cells are available. Figure 3.7 displays the flowchart of the land allocation for
AGRO.
The amount of a crop that can be produced on a micro-level grid cell is determined
by the local production P . The production P of crop type c at simulation step t for a
particular grid cell, is defined as
Pc (t) = basec × yc (t) × (1 + techc (t)) × ac (t)
Pc (t)
basec
yc (t)
techc (t)
ac (t)
(3.3)
cell level production of crop type c in time step t [Mg],
management parameter for crop type c [-],
cell level yield for crop type c in time step t [Mg km−2 ],
yield increase due to technological change for crop type c in time step t [-],
available cell area for production of crop type c in time step t [km2 ].
The crop production P is computed by combining state variables from different spatial scale levels (crop yield, crop yield changes) and the cell area a that is not required
for housing and infrastructure. The management factor base is a proxy for agricultural
28
3.3. Details
Fig. 3.7.: Flowchart of the functional part land allocation for AGRO.
management factors, which are not taken into account by LandSHIFT (see above). The
local crop yield y is updated in each time step by the Biophysic module. If not enough
(suitable) land is available to meet the country-level crop demands, unmet demands are
documented in a log-file. In case that more cropland is available as required, the land-use
type of dispensable cells is converted to “set-aside”, equivalent to fallow.
GRAZE The sub-module GRAZE accounts for the spatial and temporal dynamics of
rangeland and micro-level stocking densities. Changes in quantity and location of rangeland,
and changes in the intensity of use, are induced by alterations in the number of ruminants
(given in livestock units (LU)): sheep and goats in the regional-scale LandSHIFT version
and additionally cattle in the global-scale LandSHIFT version. Based on this information,
the demand processing calculates the yearly amount of required livestock forage, which
has to be provided by rangeland (i.e. which is covered by grazing. This is done under
consideration of the daily feed demand per LU, derived from literature, and the share of
29
3. The integrated modeling system LandSHIFT
Fig. 3.8.: Flowchart of the functional part land allocation for GRAZE.
grazing in fodder composition. The latter is derived from literature in case of the regionalscale LandSHIFT version or calculated in the initialization step in case of the global-scale
LandSHIFT version (see section 3.3.1). The residual fodder, covered by crops and crop
residues, is considered indirectly in both LandSHIFT versions.
The suitability factors taking into account in for preference ranking are NPP of rangeland
and natural vegetation, terrain slope, proximity to surface water, proximity to urban area,
and proximity to existing rangeland (Wint et al., 2003). For all suitability values except for
terrain slope, the value functions define a positive relationship. The proximity to urban area
accounts for a better market access and the proximity to cropland considers the availability
of crop residues as additional feed resource on cropland. The proximity to cropland is
defined as a Boolean relationship in terms of a first order Moore neighborhood (Fig. 3.5).
Within land allocation for GRAZE, the local forage production is calculated under consideration of the fraction of biomass digestible for ruminants. This value is derived either from
literature (e.g. Stéphenne and Lambin (2001)) or adjusted according to expert knowledge.
The biomass production, available as forage is then calculated as the NPP of rangeland
30
3.3. Details
Tab. 3.4.: Mean annual precipitation categories, applied in the regional version of LandSHIFT.
Precipitation category
Arid
Semiarid
Dry Mediterranean
Typical Mediterranean
Mesic Mediterranean
Abbreviation
AR
SA
DM
TM
MM
Mean annual precipitation
(mm)
≥ 80 to < 200
≥ 200 to < 400
≥ 400 to < 500
≥ 500 to < 700
≥ 700 to < 960
on the grid cell multiplied with the fraction of grid cell area not used for housing and
infrastructure and the fraction of digestible biomass. If the rangeland area produces more
biomass as required to meet the forage demand, the stocking density is reduced proportionately. In case, the forage production on the existing rangeland area does not fulfill
the demand, additional area has to be converted to rangeland, according to the suitability
ranking, until the demand is met.
GRAZE in the regional-scale LandSHIFT version The land allocation of GRAZE
in the regional-scale LandSHIFT version was enhanced and now implements a feedback
mechanism between grazing intensity and local biomass productivity. The new version
is based on non-linear correlation functions (see appendix A) between stocking density
(in goats and/or sheep per hectare) and biomass productivity (in tonnes per hectare),
calculated with WADISCAPE (see section 3.3.3). The correlation functions were generated
for five terrain slope categories (0◦ to <5◦ , ≥5◦ to <12.5◦ , ≥12.5◦ to <17.5◦ , ≥17.5◦
to <25◦ , ≥25◦ ) and five categories of mean annual precipitation (Tab. 3.4). Areas with
mean annual precipitation values, that are not covered by the five categories (i.e. that are
below the threshold value of 80 mm), are not considered in the simulations. Except for
the cells that feature a mean annual precipitation that is not covered by the WADISCAPE
simulations, each micro-level grid cell is attributed to one of the correlation functions
depending on the local terrain slope and mean annual precipitation.
The new allocation routine provides two opportunities to calculate the initial distribution
of rangeland and the related stocking densities:
1. The forage demand is allocated to the best suited micro-level grid cells and the landuse type of these cells is converted to rangeland. The local biomass productivity
is calculated from the nonlinear correlation function, valid for the respective cells,
assuming no former use as rangeland for these grid cells. Based on the available
biomass productivity, the local stocking density is calculated under consideration of
the forage demand per sheep and goat.
2. Based on historical information on rangeland area, the rangeland area for the corresponding year is allocated to the best suited micro-level grid cells. The land-use type
31
3. The integrated modeling system LandSHIFT
of these cells is converted to rangeland. The potential total biomass productivity on
these grid cells is calculated with the nonlinear correlation functions, assuming no
former use as rangeland for these grid cells. Based on the potential biomass productivity, the stocking density rate is adjusted and assigned to the grid cells, in order to
meet the required forage demand.
In order to calculate the local biomass productivity in the following simulation steps,
the cell’s correlation function is chosen and combined with the stocking density in the
initial simulation step. The actual local stocking density in is then calculated from this
productivity via the feed demand, and subsequently assigned to the grid cell. In the next
simulation step, this stocking density is again used to derive the new local productivity from
the cell specific correlation function. This procedure is repeated for each simulation step.
An important effect of this feedback between stocking density and biomass productivity
is the resulting self-regulation: the allocation of high stocking densities in one simulation
step results in lower biomass productivities in the next time step, in turn resulting to lower
stocking densities. In addition to the dynamic calculation of local landscape productivity,
a cell specific factor on change in biomass productivity due to climate change, also derived
from WADISCAPE simulation results, can be considered.
In addition to the newly implemented feedback mechanism, the new version of GRAZE
provides two different allocation modes, representing (a) sustainable and (b) intensive
rangeland management. These allocation modes are based on information on cell specific
stocking capacities, generated also with WADISCAPE. The allocation modes differ for the
case that the calculated stocking density exceeds the local stocking capacity, i.e., in the
case of overgrazing. Within the “sustainable” allocation mode, the local stocking capacity
defines the maximum stocking density at a grid cell. Each time the stocking density,
assessed via the local biomass productivity, exceeds the stocking capacity of the grid cell,
the stocking density is set back to the stocking capacity. Within the “intensive” allocation
mode, this limitation is not applied and the stocking density is solely limited via the local
biomass productivity. In general, the maximum stocking density for both allocation modes
is limited to 10 goats and/or sheep per hectare, which corresponds to the range of the
WADISCAPE calculations.
Furthermore, the sub-module GRAZE was extended by a reduction factor that allows
to account for an irreversible reduction of the local biomass productivity. The factor takes
action in case of overgrazing: in each simulation step, in which the allocated stocking
density exceeds the stocking capacity, the potential local biomass productivity is reduced
by a pre-defined rate. This reduction becomes operative in the following simulation step
and affects the amount the new local biomass productivity. If the local biomass productivity
falls below a threshold value, the cell becomes unsuitable for rangeland in the following
simulation steps (see chapter 4).
32
4. Modeling the feedback between
stocking density and biomass
productivity
This chapter1 describes the initial application of a simulation method that allows investigating the feedback between different rangeland management strategies and biomass
productivity and the resulting impact on land-use and land-cover change. In this initial
application, we analyzed the effects of two management strategies, sustainable and intensive, on land-use and land-cover change in the Jordan River region. For this purpose, we
used the regional version of the spatially explicit, integrated modeling system LandSHIFT.
The results of the simulation experiment indicate a strong effect of the modeled feedback
mechanism on the spatial extent of rangeland. Consequently, the results underline that the
inclusion of such feedback mechanisms in land-use change models can help to represent
and analyze the complex interactions between humans and the environment in a more
differentiated and realistic way, but they also identify the need for more detailed empirical data on rangeland degradation in order to enhance the explanatory power of land-use
change models.
4.1. Motivation
The eastern Mediterranean ecosystems pertain to the class of dryland systems and are
therefore potentially prone to desertification. The United Nations Convention to Combat Desertification defines desertification as “land degradation in arid, semi-arid and dry
sub-humid areas resulting from various factors, including climatic variations and human activities” (UNCCD, 1994). The causes of desertification are still under discussion but Geist
and Lambin (2004) identify cropland expansion, overgrazing, and infrastructure expansion
as proximate causes of desertification driven by climatic and economic factors, institutions,
national policies, population growth, and remote influences.
Current results from the Fourth Assessment Report of the Intergovernmental Panel
on Climate Change (IPCC) project increasing mean annual temperatures and decreasing
precipitation amounts accompanied by a very likely increase in length and frequency of dry
1
Based on: Koch, J., Schaldach, R., Köchy, M. (2008). Modeling the impacts of grazing land management
on land-use change for the Jordan River region. Global and Planetary Change 64: 177-187.
33
4. Modeling the feedback between stocking density and biomass productivity
spells for the Mediterranean region (Christensen et al., 2007). These changes in climate
and the projected high population growth rates (FAO, 2008) may put additional pressure
on the ecosystems of the region and presumably aggravate the desertification risk.
One option to avoid or at least diminish dryland degradation is to improve agricultural
practices towards sustainable management (Millennium Ecosystem Assessment, 2005a). In
this context, the management of grazing systems deserves special attention. Consistently
high stocking densities are identified as a cause of changes of vegetation cover/composition
(Gillson and Hoffman, 2007) and soil degradation (Ibáñez et al., 2007). These effects
can lead to a reduction of productivity of forage grasses (van de Koppel et al., 2002)
that in consequence can endanger the regional livestock production systems and human
food security. Moreover, the studies of Alados et al. (2004) and Alhamad (2006) show
that intensive grazing has negative impacts on the biodiversity of Mediterranean grassland
ecosystems.
In this chapter, we describe the initial application of a newly developed simulation based
method that uses the dynamic, spatially explicit modeling system LandSHIFT to investigate
the feedback between different rangeland management strategies and biomass productivity
and the resulting impact on land-use and land-cover change. We present the results of
a first application of this method, which distinguishes between intensive and sustainable
rangeland management, for the Jordan River region. Furthermore, in a sensitivity study
we explore the model dynamics under different assumptions regarding the reversibility of
grazing related vegetation changes and their effects on productivity (Cingolani et al., 2005).
4.2. Simulation experiment
In order to demonstrate the effects of the two different rangeland management strategies
(see chapter 3) on land-use and land-cover change, we performed a simulation experiment
for the Jordan River region. Altogether, we conducted five simulation runs based on the
Millennium Ecosystem Assessment (MEA) scenario Order from Strength. One simulation
run was carried out with sustainable rangeland management and the other four simulation runs were executed with intensive rangeland management. The simulation runs with
intensive rangeland management base on different assumptions about the effects of overgrazing on the productivity of (semi-)natural vegetation. One run represents a very resilient
grazing system, in which overgrazing does not cause irreversible productivity reduction. In
contrast, the other simulation runs represent grazing systems, in which overgrazing leads
to a decrease of productivity due to changes of vegetation cover and composition, as well
as soil degradation, which is not reversible within the simulation period of 50 years (van
de Koppel and Rietkerk, 2000; Ibáñez et al., 2007; Köchy et al., 2008). This behavior is
modeled by specifying the reduction of biomass productivity separately for each simulation
run. In absence of empirical data on rangeland degradation rates for our study region, we
performed simulation runs with 10 %, 20 % and 30 % reduction of landscape productivity
within each 5-year time step and used them to analyze the sensitivity of our model to this
34
4.3. Study region
Tab. 4.1.: Share in land area of the major land-use categories for Israel, Jordan, and the
Palestinian National Authority (PA) for the year 2000 (FAO, 2008).
Share in land area (%)
Arable land
Permanent cropland
Permanent meadows and pastures
Israel
15.62
3.97
6.56
Jordan
2.15
1.00
8.96
PA
17.61
19.93
24.92
type of degradation. While the 10 % reduction rate (equaling 2% per year) is based on
the degradation factors discussed by Stéphenne and Lambin (2001) for grazing systems
in Burkina Faso, the other rates are hypothetical values in order to illustrate the resulting
model dynamics under extreme assumptions.
The sub-module GRAZE in LandSHIFT.R exclusively considers goats and sheep, with
a feed demand of 1.35 kg dry matter per animal and day (Perevolotsky et al., 1998), of
which a fraction of 40 % is covered by grazing (Nordblom et al., 1997). The simulated
land-use and land-cover change scenarios cover a period of 50 years, from the base year
2000 up to the year 2050, in 5-year time steps. Simulations for the base year also serve for
the initialization of the model.
4.3. Study region
Our simulation experiment covers the territories of Israel, Jordan, and the Palestinian
National Authority (PA). Altogether, the land area adds up to about 116 000 km2 (Israel
about 21 000 km2 , Jordan about 89 000 km2 , and PA about 6 000 km2 ). The Palestinian
National Authority is subdivided into the Gaza Strip and the West Bank. The shares of
arable land, permanent cropland, and permanent meadows and pastures in the land areas
for the year 2000 are listed in Tab. 4.1. The study region ranges from 33.38◦ N, 34.22◦ E to
29.19◦ N, 39.30◦ E and is located in the Middle East with Israel and the Gaza Strip bordering
the eastern part of the Mediterranean Sea. The terrain in the region is structured by the
Great Rift Valley separating East and West Bank of the Jordan River. Israel is subdivided
into the low coastal plain, the central mountainous region and the southern Negev desert
(CIA, 2008). The Gaza Strip is located in the low coastal plain whereas the terrain of the
West Bank is rugged, dissected upland. A desert plateau forms the eastern part of Jordan;
in contrast the western part can be described as highland area. The lowest point of the
study region is the Dead Sea with -408 m above sea level, the highest point is the Jabal
Ram in Jordan with 1 734 m above sea level.
According to the Köppen-Geiger climate classification (Peel et al., 2007), the climate
in the region can be roughly described as follows: From northwest to southeast there is
a climate gradient from a temperate climate with hot and dry summers (northern part of
Israel and the West Bank) to arid hot desert (north-eastern part of Jordan and southern
parts of Jordan and Israel). The climate in the Gaza Strip, the western middle part of
35
4. Modeling the feedback between stocking density and biomass productivity
Israel, and a part of the north of Jordan is classified as arid hot steppe. The eastern middle
part of Israel and the western middle part of Jordan have an arid cold steppe climate, and
the eastern middle part of Jordan has an arid cold desert climate. The spatial distribution
of mean annual temperature and precipitation for the climate normal period 1961-1990
(Mitchell and Jones, 2005) is displayed in Fig. 4.1. Apart from limited natural freshwater
resources, current environmental issues in the region are, amongst others, desertification
and overgrazing (Abahussain et al., 2002; Ministry of Environment of The Hashemite
Kingdom of Jordan, 2007).
Fig. 4.1.: Spatial distribution of mean annual temperature (MAT) and mean annual precipitation (MAP) in the Jordan River region for the climate normal period 1961-1990
(Mitchell and Jones, 2005).
4.4. Materials and methods
4.4.1. Input data
The starting configuration of land-use and land-cover types is based on the IGBP Land Cover Classification data set from the Global Land Cover Characterization data base (Loveland
et al., 2000), derived from AVHRR source imagery dates from April 1992 through March
1993. The mapping of land-use and land-cover types of LandSHIFT.R onto the IGBP Land
Cover Classification is listed in Tab. 4.2. The LandSHIFT.R routines for preference ranking and land allocation use micro-level information on landscape characteristics, land-use
constraints, and land-use related model variables. We applied the GIS software ArcGIS to
extract the information from existing global data sets and mapped them geographically
onto the micro level grid cells. Human population density was derived from the Global
Rural Urban Mapping Project - GRUMP alpha (Center for International Earth Science
36
4.4. Materials and methods
Tab. 4.2.: Land-use and land-cover types implemented in LandSHIFT.R and mapping scheme for relating these to the IGBP Land Cover Legend. Only the categories of
the IGBP Land Cover Legend that occur within the study region are considered.
LandSHIFT.R
Forest
IGBP Land Cover Legend
Evergreen needleleaf forest
Deciduous broadleaf forest
Mixed forest
Open shrublands
Open shrublands
Woody savannas
Woody savannas
Grasslands
Grasslands
Permanent wetlands
Permanent wetlands
Cereals
Fruits
Vegetables
Other crops
Croplands
Urban and built-up
Urban and built-up
Cropland/Natural vegetation mosaic
Cropland/Natural vegetation mosaic
Barren or sparsely vegetated
Barren or sparsely vegetated
Rangeland
-
Fallow
-
Information Network et al., 2004), slope data is based on the HYDRO1k data set (USGS,
1998). The river network density was calculated as the line density per grid cell, based on
the HYDRO1k data set on streams (USGS, 1998). The information on infrastructure was
assessed via the VMAP0 data set on roads (NIMA, 1997). In order to derive information
on land-use constraints, we mapped data sets on national or international conservation
areas (WDPA Consortium, 2004) onto the micro-level grid cells. Moreover, the data set on
potential rain-fed and irrigated wheat yields generated with the DAYCENT model, which is
available at a spatial resolution of 30 arc minutes (Stehfest et al., 2007), was geographically
mapped onto the micro level grid cells.
The enhanced sub-module GRAZE works on a set of 25 nonlinear correlation functions,
describing the relationship between stocking density and biomass productivity (see chapter
3). The correlation functions were calculated for five precipitation categories and five slope
categories by the WADISCAPE model. The functions were separated into sections and
fitted by linear regression. Each grid cell of the study region was geographically mapped
onto a precipitation category and slope category to relate it to the specific correlation
function. The spatial distribution of the precipitation category was kept static for the entire
simulation period. The information on change in NPP of rangeland and natural vegetation
and on the changing stocking capacity, both due to a changing climate, was calculated
from the raster maps for the years 2000 and 2050 on productivity without grazing and on
stocking capacity, assuming a linear transition.
37
4. Modeling the feedback between stocking density and biomass productivity
For the base year allocation step, we regionalized the three year average production value
(1999-2001) for the crop classes cereals, fruits and vegetables (including melons) provided
by FAOSTAT (FAO, 2008). Table B.1 in appendix B summarizes the data requirements of
the simulation experiment.
4.4.2. Scenario description
In this simulation experiment, we used data from the MEA scenario Order from Strength.
This scenario depicts a “regionalized and fragmented world concerned with security and
protection” (Millennium Ecosystem Assessment, 2005b). The Order from Strength scenario
shows the highest population growth rates and the lowest economic growth rates of all
MEA scenarios. We applied this scenario to be consistent with the WADISCAPE simulation
results, which were produced for the SRES A2 scenario (Houghton et al., 2001).
The information on livestock numbers, change in crop production, and yield increase due
to technological change was derived from IMPACT simulation results (Rosegrant et al.,
2002b, 2005b). Since the MEA scenario presupposes 1997 as base year, the IMPACT data
had to be further processed for this simulation experiment. First, we used FAOSTAT values
for human population, livestock number (goats and sheep), and crop production (FAO,
2008) to calculate 3-year average values (1999-2001). Second, we identified the trends
for these parameters as calculated by IMPACT compared to the year 2000. Thereafter,
we applied these trends to the average values to calculate the parameters’ development.
Since PA is not explicitly considered in IMPACT, we applied the same trends as for Jordan,
implying that PA also belongs to the “Other West Asia and North Africa” region.
In order to derive a spatial yield distribution for the three crop classes, we scaled the
spatial information on wheat yields generated with the DAYCENT model (Stehfest et al.,
2007), with the IMPACT yields for wheat, all cereals, (sub-)tropical and temperate fruits,
and vegetables (processed for the year 2000). Yield changes are based on IMPACT results,
an additional climate effect was not taken into account. We applied a factor of 0.125
(Seré and Steinfeld, 1996) to convert the number of goats and sheep into livestock units.
Additionally, we used a regional factor of 0.8 for Israel and 0.42 for Jordan and PA, which
accounts for the regional variability of animal body sizes (Seré and Steinfeld, 1996). This
resulted in a conversion factor of 0.1 LU for one sheep or goat in Israel and 0.05 LU for one
sheep or goat in Jordan and PA. To maintain consistency with the WADISCAPE model
results, we doubled the stocking density value when assessing the productivity for Jordan
and PA from the correlation functions. Table 4.3 provides a compilation of the processed
main driving forces of LandSHIFT.R.
4.4.3. Validation
Validation is an important part of the development process of simulation models. Furthermore, validation is essential to achieve credibility in the user community (Rykiel Jr.,
1996). In case of spatially explicit simulation models of land-use change, the evaluation of
38
4.5. Results
Tab. 4.3.: Processed input to LandSHIFT.R for the Millennium Ecosystem Assessment
scenario Order from Strength, for Israel, Jordan and the Palestinian National
Authority (PA). Changes in production and yield are calculated referring to the
base year 2000.
Country
Year
Human
population
Livestock
number
(LU)
Israel
2000
6 082 667
45 220
Israel
2025
9 166 167
62 485
Israel
2050 10 704 371
73 236
Jordan
2000
4 805 333
114 064
Jordan
2025
8 399 265
253 908
Jordan
2050 10 583 197
329 322
PA
2000
3 150 333
45 566
PA
2025
5 506 483
107 134
PA
2050
6 938 249
138 954
Cer = Cereals, Fru = Fruits, Veg = Vegetables
Production change
(%)
Cer Fru
Veg
0
0
0
31
39
51
51
79
111
0
0
0
54
77
48
101 166
108
0
0
0
54
77
48
101 166
108
Yield change
(%)
Cer Fru Veg
0
0
0
26
13
7
34
21
12
0
0
0
34
34
26
52
58
50
0
0
0
34
34
26
52
58
50
predictive performance refers to both location and quantity of change (Pontius Jr., 2002).
A typical procedure to evaluate the predictive performance regarding location is to compare a simulated map with a reference map (e.g. Pontius Jr. et al. (2004)). In addition
to the data set used to calibrate the model, this approach requires a second, statistically
independent data set. For this simulation experiment, as for many others, there is a lack of
data sets that comply with this requirement. In order to validate the quantity of change,
we used FAO statistical data on permanent meadows and pastures for the years 2000 and
2005 (FAO, 2008).
4.5. Results
Figure 4.2 shows the land-use and land-cover maps for the simulation runs with sustainable
rangeland management, intensive rangeland management without productivity reduction,
and with a productivity reduction of 10 %, for the years 2000, 2025, and 2050. For visualization purposes, we aggregated the land-use types cereals, fruits, vegetables, and other
crops to the category arable land and the land-cover types forest, natural vegetation mosaic/cropland, shrub land, grassland, woody savannah, and wetland to the category natural
vegetation. The rangeland allocation algorithm applied for the base year 2000 is identical
for both rangeland management strategies, resulting in a consistent base year distribution
of land-use and land-cover types for all five simulation runs. The base year distribution
shows an extent of urban and built-up area of 962 km2 in Israel, 550 km2 in Jordan, and
409 km2 in PA. The extent of arable land in 2000 was assessed as 2 778 km2 in Israel, 2 539
km2 in Jordan, and 1 602 km2 in PA. The extent of rangeland for the base year is 398 km2
in Israel, 1 104 km2 in Jordan, and 363 km2 in PA. The residual land area is covered with
39
4. Modeling the feedback between stocking density and biomass productivity
Land-use and land-cover distribution for the year 2000
33°N
39°E
32°N
38°E
Rangeland
31°N
37°E
33°N
36°E
32°N
35°E
Natural vegetation
Urban and built-up area
29°N
00
35°E
36°E
50
50 100
100
37°E
38°E
30°N
30°N
±
(km)
(km)
200
200
Barren land
29°N
31°N
Arable land
39°E
Land-use and land-cover distribution for the year 2025
Intensive
37°E
38°E
Intensive 10 %
39°E
35°E
39°E
36°E
37°E
38°E
39°E
0
35°E
36°E
50
37°E
100
38°E
(km)
200
32°N
30°N
31°N
31°N
31°N
32°N
32°N
33°N
38°E
33°N
33°N
37°E
33°N
30°N
(km)
200
36°E
30°N
30°N
32°N
36°E
100
30°N
31°N
35°E
50
29°N
0
29°N
29°N
30°N
31°N
32°N
33°N
35°E
29°N
29°N
39°E
32°N
38°E
31°N
37°E
33°N
36°E
0
35°E
39°E
36°E
50
37°E
100
(km)
200
38°E
39°E
38°E
39°E
29°N
Sustainable
35°E
Land-use and land-cover distribution for the year 2050
36°E
37°E
38°E
39°E
36°E
37°E
100
38°E
(km)
200
39°E
37°E
31°N
32°N
33°N
36°E
30°N
33°N
32°N
31°N
35°E
50
30°N
0
30°N
30°N
35°E
0
35°E
36°E
37°E
50
100
38°E
(km)
200
29°N
39°E
33°N
38°E
29°N
30°N
(km)
200
29°N
100
37°E
29°N
32°N
35°E
50
29°N
31°N
32°N
31°N
30°N
29°N
0
36°E
32°N
35°E
33°N
39°E
33°N
38°E
32°N
37°E
Intensive 10 %
31°N
36°E
33°N
35°E
Intensive
31°N
Sustainable
39°E
Fig. 4.2.: Land-use and land-cover maps for the simulation runs with sustainable rangeland
management, intensive rangeland management without productivity reduction in
case of overgrazing, and intensive rangeland management with a productivity
reduction of 10 % in case of overgrazing.
40
4.5. Results
natural vegetation or barren land.
4.5.1. Simulation of land-use and land-cover change
Since the land-use activity grazing is in its economic importance subordinate to the landuse activities housing and infrastructure, and crop cultivation, the land-use and land-cover
changes regarding urban and built-up area and arable land are equal for all five simulation
runs. The extent of urban and built-up area for all three countries increased from approximately 1 920 km2 in 2000 by 32 % up to 2025 and 53 % up to 2050, the extent of arable
land for the study region increased from approximately 6 920 km2 in the base year by 11 %
up to 2025 and 23 % up to 2050 (Fig. 4.3).
Fig. 4.3.: Area development of (a) urban and built-up area and (b) arable land for Israel,
Jordan, and the Palestinian National Authority (PA).
Table 4.4 summarizes the extent of allocated rangeland for the three considered countries
and the years 2000, 2025, and 2050. Figure 4.4 combines the additional area demand for
rangeland compared with the year 2000 for the five simulation runs that were carried
out. The result for all simulation runs indicate an unmet feed demand for PA, associated
with an almost complete utilization of natural vegetation for grazing purposes. Figure 4.5
demonstrates the development of the unmet feed demand for PA.
4.5.2. Validation results
Figure 4.6 shows a comparison of the FAOSTAT data for the category permanent meadows
and pastures with the simulated rangeland areas for the base year 2000 and the year 2005.
41
4. Modeling the feedback between stocking density and biomass productivity
Tab. 4.4.: Simulated rangeland extent for the five simulation runs, for Israel, Jordan, and
the Palestinian National Authority (PA).
Country
Israel
Israel
Israel
Jordan
Jordan
Jordan
PA
PA
PA
Year
Sustainable
Intensive
2000
2025
2050
2000
2025
2050
2000
2025
2050
(km2 )
398
1 329
1 666
1 104
9 510
14 403
363
2 662
3 210
(km2 )
398
940
1 125
1 104
6 204
10 116
363
1 915
3 196
Intensive
10 %
(km2 )
398
1 061
1 541
1 104
6 316
10 643
363
1 901
3 194
Intensive
20 %
(km2 )
398
1 207
1 529
1 104
6 552
12 524
363
1 973
2 831
Intensive
30 %
(km2 )
398
1 365
1 675
1 104
6 917
15 387
363
2 098
2 051
Fig. 4.4.: Comparison of additional rangeland demand as compared to the year 2000 for
all five simulation runs.
4.6. Discussion
The results of our simulation experiment indicate that the modeled feedback mechanism
between rangeland management and landscape productivity has a strong effect on landuse and land-cover change in terms of the spatial extent of rangeland. Supposing that
overgrazing does not cause an irreversible reduction of landscape productivity, we see
that the area which is allocated to rangeland under the sustainable management strategy
42
4.6. Discussion
Fig. 4.5.: Unmet feed demand for the Palestinian National Authority.
exceeds the one allocated under intensive rangeland management by far. However, this
is achieved by higher stocking densities, which may have negative effects on biodiversity
(Alados et al., 2004; Alhamad, 2006) that are not considered in our simulation experiment.
Figure 4.4 displays the simulation results for all five simulation runs of the sensitivity
study: at the beginning, the area demand for rangeland for the simulation run with sustain-
Fig. 4.6.: Comparison of simulated rangeland area for the year 2000 (a) and for the year
2005 (b) with FAOSTAT data for the category permanent meadows and pastures.
43
4. Modeling the feedback between stocking density and biomass productivity
able rangeland management is much higher than for the four simulation runs with intensive
rangeland management. In the long run, the rangeland area demands calculated for the
simulation runs with intensive rangeland management including irreversible productivity
reduction, approach the rangeland area demand of the simulation run with sustainable
rangeland management.
This effect becomes even more apparent in Fig. 4.5, illustrating the unmet feed demand
for PA. This demand specifies the amount of forage, which is required but cannot be
provided, because no (suitable) land area is left to allocate this demand. Consequently, this
unmet demand has to be covered by additional feedstock. In 2050, all three simulations
delineate a higher unmet demand than the simulation run assuming sustainable rangeland
management. These results show that the model is sensitive to irreversible changes of
landscape productivity. Nevertheless, it has to be noted again that the applied reduction
factors are not based on empirical data from the region and therefore the model results
have a high uncertainty. Since various studies stress that irreversible degradation poses a
problem in Mediterranean grazing systems (Ibáñez et al., 2007; Köchy et al., 2008), there
is the demand for empirical research on the mechanistic and temporal dynamics of these
degradation effects. The results will help to further improve the explanatory power of our
model.
A comprehensive validation of the applied land-use model was beyond the scope of this
simulation experiment. Our efforts to evaluate the model performance concentrated on the
enhanced implementation of the land-use activity grazing. We compared the simulated area
of rangeland in the three countries of our study region with statistical data for permanent
meadows and pastures from FAO statistics (Fig. 4.6). For the base year 2000, the results
show low agreement. This is due to the fact that the base year allocation step was used to
initialize the spatial distribution of stocking densities by applying the unadjusted landscape
productivity values from WADISCAPE. Thus, all simulation runs generate the same spatial
extent for rangeland. For the year 2005, the correspondence between simulation results
and FAO data is much higher, especially for the simulation run with sustainable rangeland
management. However, LandSHIFT.R underestimates the area of rangeland as compared
to the FAO data. One reason for that mismatch is that our simulation experiment only
considers sheep and goats for which grazing is often executed on (semi-)natural vegetation
instead of permanent pastures (Perevolotsky and Landau, 1992). As a consequence, the
data for permanent meadows and pastures can be only a rough estimate of the actual
grazing area for these animals. Another source of uncertainty is the assumption on the
fraction of grazing at the feed composition that was set to 40 % according to Nordblom
et al. (1997).
In our simulation experiment we successfully applied the LandSHIFT.R model in order to
integrate human and environmental key processes of the land-use system and to combine
information on land-use (rangeland management) with satellite derived land-cover data of
the study region. Nevertheless, the model is still a simplistic look on the real world system as
important processes, such as rural-urban migration or the effects of additional requirements
for feedstock production from unmet feed demand on the regional extent of cropland or on
44
4.7. Conclusions
international trade patterns, are not included. Another limitation of our model approach
is that soil processes, which play an important role for degradation processes are not
explicitly modeled (Ibáñez et al., 2007). Regarding the modeling of feedback mechanisms,
the temporal resolution of the model also gains in importance. Currently, we assume that
the decision making for the allocation of stocking densities is done in 5-year time steps. Inbetween these intervals the rangeland management is assumed constant. Further research
should analyze the influence of changes in temporal resolution on the model results.
4.7. Conclusions
Result of our ongoing work is a new sub-module for the LandSHIFT.R model that enables
the simulation of feedback effects between human decision making (in form of rangeland
management strategies) and the productivity of grazing systems. In an initial simulation
experiment we could demonstrate that this type of feedback has a strong effect on the
simulated land-use pattern and the spatial extent of rangeland. Based on these results, our
research efforts will concentrate on the further refinement of the modeled decision making
processes, model validation, and on the conception of a more detailed and integrative study
design. This includes the assessment of impacts of stocking intensities on biodiversity as
well as the incorporation of more empirical data on degradation processes caused by grazing,
when they become available.
45
5. Quantifying the environmental
impact of grazing in Jordan
The regional-scale LandSHIFT version allows the simulation of rangeland management
strategies differing in the maximum applied stocking densities. The objective of the simulation experiment presented in this chapter1 was to analyze the differences in land-use
and land-cover change resulting from the application of different management strategies
in order to enhance the simulation results of LandSHIFT.R for subsequent environmental
impact studies. Therefore, the results of two simulation runs, one with sustainable and one
with intensive rangeland management were analyzed. Both simulation runs were carried out
for Jordan and differ only in the applied rangeland management strategy. The simulations
cover a period of 25 years, ranging from 2000 to 2025. Since this simulation experiment
focuses on changes regarding rangeland, the livestock number is the only driving variable
that varies over the simulation period. The development of livestock numbers over the
simulation period is based on calculations of the economic equilibrium model IMPACT on
livestock water demand under a business as usual scenario.
The simulation results were analyzed regarding the differences in the extent of rangeland
and in the spatial distribution of the relative human appropriation of net primary production
(relative HANPP). In addition, the simulation results were evaluated with a set of landscape
metrics in order to extract the impact of the rangeland management strategies on landscape
patterns. The comparative analysis of the simulation results for the year 2025 revealed
differences arising from the application of varying rangeland management strategies. The
application of the intensive management strategy resulted in a rangeland extent of 9 621
km2 . In contrast, the application of the sustainable management strategy resulted in a
rangeland extent of 12 927 km2 . The different intensity levels of resource use are reflected
in the mean relative HANPP values for rangeland: in 2025, mean relative HANPP values
are 58 % for intensive management und 47 % for sustainable management. The evaluated
landscape metrics indicated a higher value of landscape fragmentation for the simulations
with intensive rangeland management.
The sustainable rangeland management strategy is characterized by a less intensive use
of local resources associated with a higher area demand. In contrast, the application of
intensive rangeland management results in a lower area demand at the expense of the local
1
Based on: Koch, J., Schaldach, R., Kölking, C. (2009). Modelling the impact of rangeland management
strategies on (semi-)natural vegetation in Jordan. In Anderssen, R.S., R.D. Braddock and L.T.H.
Newham (eds) 18th World IMACS Congress and MODSIM09 International Congress on Modelling
and Simulation.
47
5. Quantifying the environmental impact of grazing in Jordan
resources. The relative human appropriation of net primary production considers this effect
and is therefore more suitable to evaluate the grazing intensity than a measure that includes
only the fraction of the potential natural vegetation used directly as forage. To conclude,
the enhancement of the LandSHIFT.R output with raster maps of relative HANPP and the
evaluation with landscape metrics seems suitable to enhance the value of LandSHIFT.R
results for environmental impact assessments or studies on biodiversity response.
5.1. Motivation
For terrestrial ecosystems, land-use and land-cover change is a major driver of biodiversity
change (Sala et al., 2000). The conversion of land strongly influences biodiversity via related
habitat loss and habitat fragmentation (e.g. Cushman (2006); Kruess and Tscharntke
(1994)). Besides the conversion of land, changes in the intensity of land use also exhibit a
strong influence on biodiversity or species abundance (e.g. Oehl et al. (2003); Zechmeister
and Moser (2001)). Therefore it is important to consider the intensity of land use in
models developed to generate scenarios of land-use and land-cover change. This is an
essential prerequisite for the development of scenarios including environmental policies.
The regional land-use change model LandSHIFT.R (Koch et al., 2008) implements the
simulation of grazing with different rangeland management strategies. The objective of this
simulation experiment was to quantify and visualize the impact of two different rangeland
management strategies and the resulting grazing intensities on (semi-)natural vegetation,
in order to make LandSHIFT.R results more valuable for environmental impact assessments
or studies on biodiversity response to land-use change. For this reason, LandSHIFT.R was
applied to simulate changes in areal extent of rangelands in Jordan and the associated
grazing intensity under a business as usual (BAU) scenario. Two simulation runs were
carried out, one with sustainable rangeland management and one with intensive rangeland
management. The model output of LandSHIFT.R, which comprises maps on the dominant
land-use type as well as a set of indicators representing the area loss due to land-use and
land-cover change, was extended by maps of the relative HANPP. To extract the differences
resulting from various grazing intensities, the results were compared with respect to the
areal extent and the relative HANPP of rangeland. In addition, a set of landscape metrics
was calculated in order to analyze the impact of different rangeland management strategies
on landscape patterns.
5.2. Simulation experiment
Two simulation runs were carried out, one with intensive rangeland management and one
with sustainable rangeland management. The simulations operate on a 5-year simulation
step for a period of 25 years, ranging from 2000 to 2025. The spatial resolution of the
output maps is 1 km. In 2000, the population of Jordan numbered 5 007 330 inhabitants.
48
5.3. Study region
About 232 338 tonnes of fruits, 923 537 tonnes of vegetables, and 43 838 tonnes of cereals
were produced domestically. Altogether, sheep and goat stocks in Jordan added up to
2 172 638 (specifications on population, crop production, and livestock are 3-year averages
from 1999-2001 given by FAO (2009)). Due to the fact that LandSHIFT.R considers goats
and sheep exclusively, the rangeland extent in the base year was calculated as the area
for “Permanent meadows and pastures” derived from FAOSTAT, corrected by the share of
goats and sheep in the total number of range herbivores. Based on landscape productivity,
forage demand, and rangeland management strategy, LandSHIFT.R calculated the local
stocking densities for the base year.
Since the focus of this simulation experiment is on grazing, the areal extent of all landuse activities except for grazing remains static on the year 2000 level, i.e., the livestock
number is the only driving variable that changes over the simulation period. The development of livestock numbers over the simulation period is based on livestock water demand
under a business as usual scenario as calculated by the economic equilibrium model IMPACT (Rosegrant et al., 2002a). In 2010 the increase in livestock water demand accounts
for 26 % compared to 2000, and in 2025 it accounts for 83 % compared to 2000. The
increase is directly transferred to the livestock numbers. Between 2000 and 2010 and accordingly 2010 and 2025, a linear increase is assumed. The conversion of statistical data
on goat and sheep stocks to LU is carried out as follows: one sheep or goat accounts for
0.125 LU (Seré and Steinfeld, 1996). In addition, a regional factor of 0.42 is applied that
considers the geographical variability in animal body size (Seré and Steinfeld, 1996). The
multiplication of the two factors results in a conversion factor of 0.05 LU, i.e., one sheep
or goat in Jordan equals 0.05 LU. In accordance with the range of the WADISCAPE (see
chapter 3) calculations, the stocking density for both rangeland management strategies in
LandSHIFT.R is limited to one LU per hectare.
The daily feed demand per goat or sheep is 1.35 kg dry matter (Perevolotsky et al., 1998)
of which 30 % are covered by grazing (Al-Jaloudy, 2001). Following de Leeuw and Tothill
(1993), the consumable fraction of forage is set to 30 % of the above-ground biomass.
5.3. Study region
The presented simulation experiment was carried out for Jordan. Jordan borders on Syria
in the north, on Iraq and Saudi Arabia in the east, and on Israel as well as the West
Bank in the west (Fig. 5.1). The country area of the Jordan territory adds up to about
89 000 km2 . The climate is characterized by hot, dry summers and cool, wet winters.
Mean annual precipitation ranges from about 120 mm in the south and south-east to
660 mm in the north-west. In 2000, the population of Jordan numbered approximately
5 million (FAO, 2009). About one fifth of the population lived in Amman, the country’s
administrative capital and largest city (United Nations, 2009b). With about 2.2 million
goats and sheep (FAO, 2009), the production of small ruminants is an important factor of
Jordan’s agricultural sector.
49
5. Quantifying the environmental impact of grazing in Jordan
Fig. 5.1.: Study region.
5.4. Materials and methods
5.4.1. Relative HANPP
In order to quantify and visualize the environmental impact of different grazing intensities, grid cells assigned as rangeland feature, in addition to the dominant land-use type,
information on the relative HANPP. The relative HANPP is derived from the HANPP
that indicates “the aggregate impact of land use on biomass” (Haberl et al., 2007b). The
HANPP is calculated as given in Haberl et al. (2007a):
(5.1)
HAN P P = ∆N P PLC + N P Ph
∆N P PLC NPP changes induced by soil degradation, soil sealing, and ecosystem change,
N P Ph
NPP harvested or destroyed during harvest.
Under consideration of equation 5.1, the relative HANPP is calculated as
HAN P P
relative HAN P P =
N P P0
N P P0
· 100[%],
(5.2)
NPP of the potential natural vegetation.
N P P0 is derived from the non-linear correlation functions between stocking density and
productivity of green biomass applying a stocking density of zero (see chapter 3). The
N P P of the actually prevailing vegetation (N P Pact ), which is determined via the applied
stocking density in the previous simulation step, is subtracted from the N P P0 in order
to assess ∆N P PLC . Since the focus of this simulation experiment is on rangeland, the
N P Ph displays the fraction of N P Pact that is used directly as forage.
50
5.4. Materials and methods
Tab. 5.1.: Classification of land-use and land-cover types for the analysis with landscape
pattern metrics.
Heavily human-influenced vegetation cover
Urban/built-up area
Fruits
Vegetables
Cereals
Other crops
Rangeland (rel. HANPP ≥ 50 %)
(Semi-)natural vegetation cover
Shrubland
Grassland
Permanent wetland
Cropland/Nat. veg. mosaic
Barren/sparsely vegetated
Rangeland (rel. HANPP < 50 %)
Water bodies
5.4.2. Landscape metrics
In order to quantify the impact of different rangeland management strategies on the change
of landscape patterns, the results of the two simulation runs were analyzed with a set of
landscape metrics. A recent overview of the use of landscape metrics and indices is given
in Uuemaa et al. (2009).
For the evaluation of the simulation results with landscape metrics at the class level,
the land-use and land-cover maps for the years 2000 and 2025 were combined with the
corresponding maps of relative HANPP. Therefore, the land-use and land-cover types were divided into the two classes “heavily human-influenced vegetation cover” and “(semi-)
natural vegetation cover” (Tab. 5.1). The latter was evaluated with a set of metrics at the
class level regarding fragmentation and connectivity of patches. In this context, a patch is
a cluster of connected cells of the same class. Two straightforward metrics at the class level
were chosen: the number of patches (NP) describes the subdivision or fragmentation and
the connectance (CONNECT) describes the connectivity of the focal class. The connectance was assessed with a threshold distance of 10 km. The calculations were carried out
with the software package FRAGSTATS. A detailed description of the software package
and the applied metrics is given in McGarigal et al. (2002).
5.4.3. Validation
A comprehensive description of the validation of LandSHIFT.R is given in (Koch et al.,
in review). Validation efforts were made for the Jordan River Region that covers Israel,
Jordan, and PA. Most attempts to validate the location of change base upon a pixel-bypixel comparison of a simulated map and a reference map of land-use change (e.g. Pontius
Jr. et al. (2004)). These attempts require a second statistically independent land-use data
set. Since such a data set is not available, the location of change is validated by testing
the plausibility of the model assumptions regarding the suitability for the three land-use
activities. This is done with a relative operating characteristic (ROC) analysis (Pontius Jr.
and Schneider, 2001). The performance measure of the ROC analysis is the area under
51
5. Quantifying the environmental impact of grazing in Jordan
curve (AUC), calculated by trapezoidal approximation. The ROC analysis was carried out
for all three land-use activities. The resulting AUC values are 0.9 for METRO, 0.74 for
AGRO, and 0.84 for GRAZE. All three AUC values are significantly higher than the value
for randomly distributed suitability values (0.5) and hence indicate that changes in land use
are found predominantly at locations where LandSHIFT.R allocates high suitability values.
5.5. Results
The simulation of land-use and land-cover change resulted in 368 km2 urban/built-up area,
an arable land extent of 2 425 km2 and 6 464 km2 rangeland for the year 2000 (Fig. 5.2). A
further expansion of urban/built-up area and arable land is not taken into account in this
simulation experiment. Hence, the respective area extents remain constant over the entire
simulation period. In 2025, the spatial extent of rangeland under intensive management
amounts to 9 621 km2 . This is an increase of 49 % compared to the rangeland extent
simulated for the year 2000. In contrast, the rangeland under sustainable management
doubled. This equates to a rangeland extent of 12 927 km2 (Fig. 5.3).
The visualization of the relative HANPP on rangeland reveals clear differences between
the two rangeland management strategies (Fig. 5.4). In 2025, the relative HANPP under
intensive rangeland management ranges from 30 % to 88 % and the relative HANPP under
sustainable rangeland management ranges from 8 % to 73 %. Mean relative HANPP values
for the year 2025 are 58 % for intensive management and 47 % for sustainable management.
It is important to keep in mind that equal relative HANPP values do not necessarily result
in equal forage production values on the respective grid cell.
Because none of the simulations for the year 2000 result in relative HANPP values equal
or higher than 50 %, the processed map for the evaluation of landscape metrics is identical
for both management strategies. The evaluation of these maps disclosed 18 patches (NP)
of the class “(semi-)natural vegetation cover“ and a connectance (CONNECT) of 20 %.
The analysis of the processed map for the year 2025 under intensive rangeland management
showed 135 patches and a connectance of 5.7 %. The analysis of the corresponding map
under sustainable management showed 111 patches and a connectance of 6.3 %.
5.6. Discussion and conclusions
The combination of land-use and land-cover maps with maps displaying the relative HANPP reveals the main difference between the two rangeland management strategies. The
sustainable management strategy is characterized by less intensive use of local resources.
However, the area demand required to fulfill the demand for forage assessed for the simulation run with sustainable management exceeds the area demand calculated for the
simulation run with intensive rangeland management. On the other hand, the pressure on
local resources is much higher under the intensive management strategy.
52
5.6. Discussion and conclusions
Fig. 5.2.: Land-use and land-cover maps for (a) 2000 as well as for 2025 with (b) intensive and (c) sustainable rangeland management. The simulated rangeland extent
for the year 2000 is equal for both simulation runs. The category “Additional
rangeland” describes the change in rangeland up to 2025 compared to 2000.
A secondary effect, which is not observable in maps of the dominant land-use or landcover type, is the reinforcing effect of high stocking densities on the productivity of (semi-)
53
5. Quantifying the environmental impact of grazing in Jordan
Fig. 5.3.: Development of urban and built-up area, arable land, and rangeland for the
simulation runs with (a) intensive and (b) sustainable rangeland management.
natural vegetation. The application of intensive rangeland management results in a larger
reduction of biomass productivity. This effect is implemented in LandSHIFT.R via the
correlation functions between stocking density and biomass productivity, calculated by the
WADISCAPE model. The relative HANPP considers this effect in terms of N P P changes
induced by applied management practices (∆N P PLC ). For this reason, the relative HANPP
is more suitable to evaluate the grazing intensity than a measure that represents solely the
fraction of the N P P0 used directly as forage, such as the stocking density.
54
5.6. Discussion and conclusions
Fig. 5.4.: Spatial distribution of relative HANPP in 2000 applying intensive (a) and sustainable (b) rangeland management and in 2025 applying intensive (c) and
sustainable (d) rangeland management.
Taking into account the relative HANPP for the analysis of land-use and land-cover
maps with landscape metrics, allows the inclusion of grazing intensities into the analysis
of landscape patterns. Under the conditions described above, NP showed a higher increase
and CONNECT showed a higher decrease for the simulation run with intensive rangeland
management as compared to the simulation run with sustainable rangeland management.
The combination of these two measures indicated a higher value of landscape fragmentation
under the application of the intensive rangeland management strategy. A difficulty of the
evaluation of landscape fragmentation with landscape metrics is the choice of an adequate
threshold value to distinguish between rangeland classified as “heavily human-influenced
55
5. Quantifying the environmental impact of grazing in Jordan
vegetation cover” and “(semi-)natural vegetation cover”. One improvement strategy could
be to derive the threshold value depending on climate conditions (e.g. precipitation), to
depict the influence of grazing on landscape productivity.
Nevertheless, the enhancement of the LandSHIFT.R output with raster maps of relative
HANPP and the evaluation with landscape metrics seems suitable to enhance LandSHIFT.R
results for environmental impact assessments or studies on biodiversity response. Since
the application of moderate grazing is discussed as one possibility to maintain the open
landscapes in the Mediterranean region (Köchy et al., 2008), the enhanced simulation
results of LandSHIFT.R could be used to support decisions on future strategies regarding
environmental policies or the assignment of conservation areas. Another possible application
is the analysis of vertical water transport since grazing intensity strongly influences the
leaf area index (Menzel et al., 2009). Moreover, the simulation results support studies of
nutrient flow with livestock as vector for organic material and flow of mineral nutrients
from and to rangelands (Bationo and Buerkert, 2001).
56
6. Future land-use and land-cover
change scenarios for the Jordan
River region
The Jordan River region has one of the lowest per capita water availabilities worldwide. In
order to develop pathways to increase the benefits from the scarce regional water resources
for both humans and ecosystems, the GLOWA Jordan River project comprises the elaboration of regional development scenarios. Since land-use and land-cover change affect water
quantity and water quality as well as biodiversity and ecosystem functioning, scenarios of
future land-use and land-cover change form an essential part of the regional development
scenarios. This chapter1 describes how the regional-scale LandSHIFT version was adjusted
to the geographic, biophysical, and socio-economic conditions of the Jordan River region
and how the model was applied as integration tool within the scenario analysis of the
GLOWA Jordan River project. Furthermore, this chapter provides a description of a set of
four comprehensive, spatially explicit land-use and land-cover change scenarios for Israel,
Jordan, and PA up to 2050. Two of the scenarios show large changes in land use and land
cover. Reasons are an increase in livestock numbers and the application of a sustainable rangeland management strategy, respectively. Moreover, the land-use and land-cover
change scenarios identify the need for the readjustment of some scenario assumptions as
well as for more detailed, spatially explicit yield information covering Israel, Jordan, and
PA.
6.1. Motivation
With total renewable water resource values ranging from 52 to 535 m3 per capita and year
(FAO, 2003), the Jordan River region (Israel, Jordan, and PA) has one of the lowest water
availabilities per capita worldwide. These are far below the threshold value of 1 000 m3
per capita and year indicating chronic water scarcity (Falkenmark and Rockström, 2004).
Current limitations of water availability are likely to aggravate in the future due to ,e.g.,
high population growth rates, economic development, and changing climate conditions.
The interdisciplinary and international research project GLOWA Jordan River (GLOWA JR)
1
Based on: Koch, J., Schaldach, R., Onigkeit, J., Ceglarek, T., Alcamo, J., Köchy, M., Wolff, H.-P., Kan,
I. (in review). Future scenarios of land-use and land-cover change for the Jordan River region. Ecology
and Society.
57
6. Future land-use and land-cover change scenarios for the Jordan River region
aims at providing scientific support for sustainable water management and the development
of strategies to increase the benefits from regional water resources for both humans and
ecosystems (http://www.glowa-jordan-river.de/). In order to support the planning of these
strategies, a scenario exercise was initiated with intensive participation of stakeholders from
the region as well as scientists of GLOWA JR.
Land-use and land-cover change (LULCC), for example in terms of conversion of natural
to agricultural ecosystems, affects water quantity as well as water quality (e.g. DeFries
and Eshleman (2004); Foley et al. (2005); Scanlon et al. (2007)). Moreover, changes in
land-use and land-cover induce habitat destruction, degradation, and fragmentation, and
thereby give rise to biodiversity loss and species extinction (e.g. Pimm and Raven (2000);
Sala et al. (2000)). Hence, possible future dynamics of regional land-use and land-cover
change form an essential part of the GLOWA JR scenario exercise.
In this chapter, we present the application of the integrated land-use change model
LandSHIFT.R within the scope of the scenario analysis of GLOWA JR, in order to develop
a set of plausible, transparent, and comprehensive LULCC scenarios for the Jordan River
region. LandSHIFT.R integrates land-use related information generated in the context of
the scenario process and by other scientific sub-projects of GLOWA JR, into spatially
explicit scenarios of LULCC. With a spatial resolution of 1 km, these results are suited
to serve as input for subsequent environmental impact studies and hydrological modeling
(e.g. Menzel et al. (2009)).
The objectives of this chapter were, on the one hand, to describe how LandSHIFT.R
was employed to support the GLOWA JR scenario analysis and, on the other hand, to
provide a description of a set of LULCC scenarios for the Jordan River region in order
to support the evaluation of impacts of different land-use and land-cover options on the
regional ecosystems and freshwater resources.
6.2. Study region
The study region is located in the Middle East and covers Israel, Jordan, and PA (Gaza
Strip and West Bank). It borders on Lebanon and Syria in the north, on Iraq and Saudi
Arabia in the east, on Egypt in the south-west, and on the Mediterranean Sea in the west
(Fig. 6.1). The study region ranges from 29.19◦ N to 33.38◦ N and from 34.22◦ E to 39.30◦ E
and covers 116 000 km2 .
The climatic conditions in the region are characterized by hot, dry summers and cool,
wet winters (Executive Action Team, 1998). Mean annual precipitation ranges from 900
mm in the northern tip of Israel to less than 100 mm in the desert areas of the study
region, which are located in the south of Israel and in the south and south-east of Jordan.
In the year 2000, the study region sustained approximately 14 million inhabitants, of
which 43 % lived in Israel, 34 % in Jordan, and 23 % in PA. About 0.3 million tonnes of
cereals, 1.8 million tonnes of fruits, and 3.0 million tonnes of vegetables (including melons)
were produced in the study region. Altogether, the study region sustained 3.7 million goats
58
6.3. Materials and methods
±
Elevation (m)
8 238
- 407
0 25 50
100
(km)
Fig. 6.1.: The study region covers Israel, Jordan and the Palestinian National Authority
(Gaza Strip and West Bank).
and sheep (FAO, 2009). The region’s largest cities are Amman, Jerusalem, Tel Aviv, and
Gaza.
6.3. Materials and methods
6.3.1. The scenario analysis in GLOWA Jordan River
Overview
For the development of the “GLOWA JR scenarios of regional development under global
change” (Anonymous, 2009) the Story and Simulation approach (SAS) was applied (Alcamo, 2001, 2008). Of these so-called “regional development scenarios” the LULCC scenarios
form an essential part. A characteristic of the SAS approach is that it combines qualitative
and quantitative elements in a balanced way, which makes the resulting scenarios suitable
for planning purposes and generally understandable at the same time. A distinctive feature of SAS is the intensive participation of local experts, involved in the management of
regional water resources. In general, SAS is an iterative procedure consisting of five steps:
1. A group of representative stakeholders and other experts (the scenario panel) drafts
qualitative scenario storylines.
2. Key driving forces of the storylines are quantified with the help of the scenario team.
59
6. Future land-use and land-cover change scenarios for the Jordan River region
3. The modeling teams use the driving forces as well as the storylines to quantify the
implications of the scenarios for land use and land cover, hydrology, and other issues.
4. The scenario panel revises the storylines based on the modeling results and other
comments. Steps 2 to 4 are repeated until the storylines and the quantifications are
acceptable for the local experts.
5. The scenarios are distributed for a general review, revised, and finally published.
The GLOWA Jordan River scenarios of regional development
The GLOWA JR scenario panel developed four scenarios for a period till 2050. The four
scenarios, named Poverty and Peace (PP), Modest Hopes (MH), Suffering of the Weak
and the Environment (SWE), and Willingness and Ability (WA), are characterized by
contrasting assumptions on economic development and the political situation, including the
corresponding mode of water sharing between Israel, Jordan, and PA, which the scenario
panel identified as the two determinants of future development in the region. In qualitative
terms and with respect to the driving forces of LULCC, the four GLOWA JR scenarios can
be described as follows:
Modest Hopes The Modest Hopes scenario assumes that no peace agreement can be
reached in the future but that economic prosperity prevails, kindled by international
donors. The economic prosperity leads to a net-immigration to the region resulting
in a medium population growth. Agricultural production on the regional scale is
dominated by Israel and remains almost constant. However, the trends differ strongly
between Jordan and PA, with high production increases in Jordan and only small
increases for PA relative to the other scenarios. The small increase in PA is due to
the relatively small population growth in PA under this scenario. The number of
sheep and goats shows a medium increase as compared to the other three scenarios.
Poverty and Peace The Poverty and Peace scenario is characterized by a peaceful development in the region, however, without economic prosperity. Although there is a
net-immigration to PA due to the peaceful conditions, the lack of economic prosperity in the region leads to a low regional population growth. The livestock numbers
(sheep and goats) are slightly more than doubling and agricultural production in
the region is slightly increasing. Nevertheless, livestock numbers and agricultural
production range at the lower end of all scenarios.
Suffering of the Weak and the Environment The Suffering of the Weak and the Environment scenario is a pessimistic scenario in which neither peace nor economic
growth can be reached. It exhibits a population growth identical to that of the PP
scenario. Due to ongoing conflicts, more people leave the region than come in but
birth rates remain high. The livestock number almost doubles but remains low relative to other scenarios. This is also true for the production of field crops: On the
60
6.3. Materials and methods
regional level, the production shows the lowest increase of all scenarios and for cereals even a decrease of production in Israel. Since production is strongly related to
population growth, the production of field crops increases in PA and Jordan.
Willingness and Ability The Willingness and Ability scenario describes a world in which
peace and economic prosperity reign. Despite decreasing birth rates, this scenario
shows the highest population growth due to immigration motivated by peaceful
conditions and economic growth. The livestock number strongly increases due to
population growth and assumptions of the associated MEA scenario on the export
of meat and other animal products. This is especially the case for Jordan with its
large area potentially available for grazing. Since enough water is available under this
scenario, the agricultural production in all countries shows the strongest increase of
all scenarios. It is also assumed that the productivity increase is the highest of all
scenarios.
The scenarios of regional development as input for LandSHIFT.R
As part of the GLOWA JR scenarios, quantitative estimates of population growth, crop production, yield increase due to technological change, and livestock numbers were generated
as driving forces for the regional land-use change model LandSHIFT.R. LandSHIFT.R combined this information with data on wheat yields calculated with DAYCENT (Parton et al.,
1998; Stehfest et al., 2007), information on biomass productivity and stocking capacity
simulated by WADISCAPE (Köchy, 2007; Köchy et al., 2008), information on landscape
characteristics (e.g. slope), land-use properties (e.g. population density), land-use constraints (e.g. conservation areas), and census data on crop production or area statistics into
scenarios of LULCC for Israel, Jordan, and PA. In appendix B, a summary of the quantitative input data for this simulation experiment is listed. Additionally, we used qualitative
information on regional planning and environmental policy from the scenario storylines to
parameterize LandSHIFT.R. For example, we assumed a sustainable rangeland management strategy for the MH scenario since the storyline for this scenario indicates a growing
importance of landscape protection against industrial agriculture (Anonymous, 2009).
In the context of the scenario analysis of GLOWA JR, quantitative estimates of driving
forces of land-use change were generated in two ways. The scenario panel described the
dynamics of driving forces such as population growth in spoken language expressions,
e.g. “population growth is medium”, based on the scenario storylines. For these qualitative
estimates each member of the scenario panel provided the related quantitative ranges which
are valid for the expression irrespective of the scenario, state or time step. These estimates
were converted to single numbers, suitable as model input, by applying a method based
on fuzzy set theory (Onigkeit et al., 2007). For livestock numbers, agricultural production
for Jordan and PA, and assumptions on yield increase due to technological change, the
scenario panel did not provide estimates. Instead, we adapted the four MEA scenarios
(Carpenter et al., 2005) by using MEA trends of per capita production together with FAO
61
6. Future land-use and land-cover change scenarios for the Jordan River region
data and the GLOWA JR assumptions on population growth. The MEA production values
are based on simulations of the agro-economic model IMPACT (Rosegrant et al., 2002b,
2005b). We give a summary of the scenario assumptions in Tab. 6.1.
Tab. 6.1.: Summary of quantified scenario assumptions for the land-use and land-cover
change scenarios.
Scenario driver
Scenario
Percentage change between 2000 and 2050
MH
PP
SWE
WA
140
130
130
170
Population
Agricultural production
Cereals
0
Fruits
30
Vegetables
47
Yield increase (Jordan and PA)
Cereals
77
Fruits
76
Vegetables
97
Livestock number
240
Rangeland management Sustainable
-14
40
43
-20
39
39
2
97
85
61
69
82
140
Intensive
51
58
65
140
Intensive
85
83
103
530
Intensive
The quantified estimates on population growth, per capita income, and water availability then served as input for the agro-economic model VALUE (Kan et al., 2007) which
simulates future crop production in Israel. Since Israel is working on a high technological
standard, we assumed no yield increase due to technological change. Figure 6.2 displays
the workflow of the simulation experiment.
6.3.2. Derivation of value functions for preference ranking
A value function serves for the transformation of the co-domain of the suitability factors
(see equation 3.2) to a co-domain from 0 to 1 (Geneletti, 2004). Altogether, we assessed
ten value functions, one for each suitability factor of each land-use activity (Tab. 6.2).
In order to derive these functions, we employed the procedure described in Heistermann
(2006). This approach bases upon two types of data sets: On the one hand, the distribution
of the respective suitability factor over the entire study region and, on the other hand, the
distribution of the respective suitability factor over the grid cells covered with the land-use
activity under consideration. For METRO we used cells classified as “Urban and built-up”
in the GLCC land-cover data set (Loveland et al., 2000) and cells featuring a population
density of greater than or equal to 5 000 inhabitants/km2 (Center for International Earth
Science Information Network et al., 2004). For AGRO we used cells classified as “Cropland”
in the GLCC data set and for GRAZE we used cells featuring a small ruminants density
higher than 200 heads/km2 (FAO, 2007).
62
6.3. Materials and methods
Fig. 6.2.: Workflow of the simulation experiment on future land-use and land-cover change
scenarios for the Jordan River region.
We divided the domain of the suitability factor for the two data sets into class ranges.
In this simulation experiment, the class ranges were defined by the unique data values.
Subsequently, we evaluated the frequency of occurrence for both data sets and constructed the corresponding cumulative frequency distributions. Third, we computed the ratio
between the frequency distribution of the suitability factor over the land-use activity and
the frequency distribution of suitability factor over the entire study region. Finally, we
derived a value function from the resulting function via a nonlinear least squares fit. We
assumed all value functions to increase monotonically, except for the value function for the
suitability factor slope, which we assumed to decrease monotonically. A monotonic increase
represents a rise in suitability with an increasing factor value and vice versa.
6.3.3. Model validation
Most attempts to validate spatially explicit land-use change models base upon a pixel-bypixel comparison of two consistent land-cover maps - a simulated one and an observed
one (Kuhnert et al., 2005; Pontius Jr. et al., 2004). In addition to the initial land-use
and land-cover map, this approach requires a second, statistically independent one with
comparable land-cover classes. Since we lack such a map, we used an alternative validation
approach: the relative operating characteristic analysis (ROC) (Pontius Jr. and Schneider,
2001).
In this simulation experiment, the ROC bases upon a suitability map (displaying the
63
6. Future land-use and land-cover change scenarios for the Jordan River region
Tab. 6.2.: Composition and parameterization of suitability factors and land-use constraints
for the land-use activities METRO, AGRO, and GRAZE.
Land-use activity
METRO
Classification
Suitability factor
Land-use constraint
AGRO
Suitability factor
Land-use constraint
GRAZE
Suitability factor
Land-use constraint
Description
Slope
Infrastructure
Land-use transition
Conservation area
Slope
Proximity to urban/built-up area
Infrastructure
Land-use transition
Conservation area
Marginal yield
NPP
Slope
Accessibility of surface water
Proximity to urban/built-up area
Proximity to arable land
Land-Use transition
Conservation area
Marginal NPP
Weight
1/2
1/2
1/3
1/3
1/3
1/5
1/5
1/5
1/5
1/5
priority in which cells are selected for LULCC) and a categorical map (displaying observed
LULCC). The categorical map was obtained by a spatial subtraction of observed raster
maps for two points in time. Grid cells that feature a change were categorized as “change”.
All other cells were categorized as “non-change”. The two maps were combined and subsequently, the observed suitability values were grouped into deciles, and the number of
“change” cells per decile was evaluated and opposed to the number of “non-change” cells
per decile, in form of a cumulative frequency distribution. The performance measure of
ROC is the area under curve (AUC), calculated by trapezoidal approximation. In case the
priority given in the suitability ranking perfectly matches the sequence in which real landcover changes occur, the AUC equals 1.0. Randomly distributed suitability values result in
an AUC of 0.5.
The temporal coverage of the available observed data does not fit the temporal coverage of the simulations. Hence, we performed ROC instead of comparing observed and
simulated changes. We performed one ROC analysis for each land-use activity. Therefore,
we compiled separate categorical maps for the land-use activities. For METRO we used
CIESIN population density maps for the years 1990 and 2000 (Center for International
Earth Science Information Network et al., 2005). We assumed cells with more than 5 000
inhabitants per square kilometer to be urban cells. For GRAZE we compiled the small
ruminant density maps adjusted to match FAO totals for the years 2000 and 2005 (FAO,
64
6.4. Simulation experiment
2007), assuming cells with more than 200 heads per square kilometer to be rangeland cells.
For AGRO, sufficient data was available only for Israel and, consequently, we implemented
the analysis of this land-use activity only for Israel. We evaluated maps on the distribution
of arable land in Israel for the years 1995 and 2002. The suitability maps for METRO,
AGRO, and GRAZE were a direct output of LandSHIFT.R. The resulting AUC values (Fig.
6.3) are significantly higher than the value for randomly distributed suitability values and
indicate that LandSHIFT.R simulates high suitability values predominantly at locations
where LULCC was observed.
Fig. 6.3.: Relative operating characteristic (ROC) curves for the three land-use activities
METRO, AGRO, and GRAZE. The gray 45◦ line indicates the ROC curve for
randomly distributed suitability values. The area under curve (AUC) is the performance measure of ROC.
6.4. Simulation experiment
In order to demonstrate the impacts of the different scenario assumptions on LULCC for
Israel, Jordan, and PA, we conducted four simulation runs - one for each scenario. The
model configurations for the respective runs only differ regarding the scenarios assumptions,
summarized in Tab. 6.1. The simulation runs cover a period of 50 years, from the base
65
6. Future land-use and land-cover change scenarios for the Jordan River region
year 2000 up to 2050 in 5-year time steps. Simulations of the base year conditions also
serve for the initialization of the model.
The basic principle of METRO is to convert the population to a cell specific population
density value. For this purpose, one part of the population is allocated to urban areas and
the other part of the population is allocated to rural areas. The fraction of population
allocated to urban areas is 65 % (Martino and Zommers, 2007). In case that the rural
population density exceeds 5 000 inhabitants/km2 , or the area demand for housing and
infrastructure on a grid cell exceeds 80 % of the grid cell size, the land-cover type of the
grid cell is changed to urban. The maximum population density per grid cell is 26 098
inhabitants/km2 , derived from the population density map for the study region for the year
2000 (Center for International Earth Science Information Network et al., 2004). The per
capita area demand is derived from (Erb et al., 2007).
For Jordan and PA, the activity AGRO comprises the land-use types cultivation of cereals, fruits, and vegetables (including melons) plus potatoes as given by the FAOSTAT
data base (FAO, 2009). The composition of considered crop categories for Israel differs
from the one for Jordan and PA, because for Israel, we used VALUE simulation results.
For Israel, LandSHIFT.R simulates the three crop categories fruits, vegetables, and field
crops. Furthermore, AGRO comprises the cumulative land-use type cultivation of “other
crops”, which is assigned to grid cells classified as cropland in the initial land-cover map
(Loveland et al., 2000), and on which LandSHIFT.R does not allocate fruits, vegetables,
or cereals/field crops in the base year simulation step. The area covered with “other crops”
remains constant in the following time steps.
The activity GRAZE comprises the land-use type rangeland. To derive the forage demand from the livestock number, we assume one sheep or goat to equal 0.125 LU. In
addition, we apply a regional factor for Israel (0.82) and Jordan/PA (0.42) that considers
the geographical variability in animal body size (Seré and Steinfeld, 1996). The daily feed
demand per goat or sheep is 1.35 kg dry matter (Perevolotsky et al., 1998) of which we
assume 30 % to be covered by grazing (Al-Jaloudy, 2001). The consumable fraction of the
above-ground green biomass is set to 75 %.
6.5. Results
For urban and built-up area (land-use activity METRO), the simulation results for all four
scenarios indicate a continuous expansion (Tab. 6.3). The results for the scenarios PP and
SWE display an increase in urban and built-up area of 1 077 km2 by 2050. The outcomes
for MH give an increase of 1 081 km2 and thus deviate only slightly from the results for
PP and SWE. By contrast, the results for the WA scenario demonstrate a considerably
larger expansion of urban and built-up area up to 2050 (1 326 km2 ). Figure 6.4 compares
the area demand for the three land-use activities as fraction of the state area for (a) Israel,
(b) Jordan, and (c) PA. Figure 6.5A shows the distribution of urban and built-up area
between the three states. The largest expansion of urban and built-up area, compared with
66
6.5. Results
base year conditions, was calculated for Jordan. The lowest expansion was calculated for
Israel. The expansion of urban and built-up area of all scenarios, accounts for merely 1 %
of the total study area (Fig. 6.6). Since total human population is the sole driving force of
change in urban and built-up area, the order of the scenarios with regard to the expansion
of urban and built-up area is geared to the assumptions on population growth (Tab. 6.1),
assuming the highest growth for WA and similar growth for all other scenarios.
Tab. 6.3.: Urban and built-up area (METRO), area covered with fruits, vegetables, cereals/field crops, and other crops (AGRO), and rangeland area (GRAZE), as well
as the changes in extent as compared to the extent in 2000 for the scenarios
Modest Hopes (MH), Poverty and Peace (PP), Suffering of the Weak and the
Environment (SWE), and Willingness and Ability (WA).
Land-use activity
Year
METRO
2000
2010
2030
2050
2000
2010
2030
2050
2000
2010
2030
2050
AGRO
GRAZE
MH
(km2 )
(%)
1 464
0
1 661
+ 14
2 007
+ 37
2 545
+ 74
7 331
0
6 804
-7
6 945
-5
8 199
+ 12
10 892
0
11 366
+4
13 736 + 26
25 844 + 137
PP
(km2 )
(%)
1 464
0
1 667 + 14
2 001 + 37
2 541 + 74
7 331
0
6 929
-6
7 095
-3
7 954
+9
10 892
0
10 706
-2
10 706
-2
10 964 + 1
SWE
(km2 )
(%)
1 464
0
1 667 + 14
2 001 + 37
2 541 + 74
7 331
0
7 124
-3
6 907
-6
8 118 + 11
10 892
0
10 706
-2
10 708
-2
10 940
0
WA
(km2 )
(%)
1 464
0
1 726
+ 18
2 210
+ 51
2 790
+ 91
7 331
0
6 944
-5
7 470
+2
9 234
+ 26
10 892
0
10 752
-1
13 315 + 22
25 179 + 131
Arable land increases up to 2050 in all four scenarios (land-use activity AGRO, Tab. 6.3).
Again, the results for the scenarios MH, PP, and SWE vary only slightly (between 623 km2
and 868 km2 additional arable land) whereas the results for WA show a noticeably higher
expansion in arable land (1 903 km2 ). For all scenarios, the largest fraction of arable land
is located in Israel, the second largest fraction in Jordan, and the least fraction in PA (Fig.
6.5B). Nevertheless, only the WA scenario results in an expansion of arable land in Israel.
In addition, the results for the WA scenario reveal the largest expansion of arable land in
all three states (Israel 4 %, Jordan 37 %, and PA 77 %). As percentage of the total study
region, the expansions of arable land for the different scenarios range from 0.5 % (PP)
to 1.6 % (WA) and thus exhibit the same order of magnitude as the expansion of urban
and built-up area (Fig. 6.6). For the Bet She’an Basin, located in Israel, the simulations
display an unmet crop demand for all scenarios. In general, the development of arable land
is geared to domestic production as given by VALUE for Israel and IMPACT for Jordan
and PA, as well as to the yield increase due to technological change (Tab. 6.1).
The rangeland simulations (land-use activity GRAZE) display a higher variability between
the four scenarios as compared to the other land-use activities. Initially, rangeland area is
decreasing for the scenarios PP, SWE, and WA, but in 2050 the simulation results reflect
67
6. Future land-use and land-cover change scenarios for the Jordan River region
Fig. 6.4.: Fraction of the land-use activities in the state areas of (A) Israel , (B) Jordan,
and (C) PA given as percentage for the base year and for the year 2050 for the
scenarios Modest Hopes (MH), Poverty and Peace (PP), Suffering of the Weak
and the Environment (SWE) and Willingness and Ability (WA). Surplus means
the remaining area with natural vegetation.
68
6.5. Results
Fig. 6.5.: Absolute area in 2000 and 2050 for the land-use activities (A) METRO, (B)
AGRO, and (C) GRAZE, for the scenarios Modest Hopes (MH), Poverty and
Peace (PP), Suffering of the Weak and the Environment (SWE), and Willingness
and Ability (WA).
69
6. Future land-use and land-cover change scenarios for the Jordan River region
Fig. 6.6.: Aggregated land-use and land-cover change trends in 2050 for the scenarios
Modest Hopes (MH), Poverty and Peace (PP), Suffering of the Weak and the
Environment (SWE), and Willingness and Ability (WA). The y-axis represents
the absolute area as percentage of the total study region area. Surplus means
the remaining area covered with natural vegetation.
an expansion of rangeland for all four scenarios (Tab. 6.3). In contrast to the other three
scenarios, the simulation results for MH show a continuous expansion of rangeland (Tab.
6.3). For all scenarios the largest fraction of rangeland is located in Jordan (Fig. 6.5C),
where GRAZE at the same time represents the most area consuming land-use activity (Fig.
6.4B). The results for PP and SWE depict only a marginal increase in rangeland area in
the study region (72 km2 for PP and 48 km2 for SWE), whereas the results for MH and
WA show a clear expansion of rangeland (14 952 km2 for MH and 14 287 km2 for WA).
The expansion for MH and WA accounts for approx. 12 % of the total study area (Fig.
6.6).
Obviously, the scenarios MH and WA stand out due to the high demand for rangeland
area. For MH, the expansion of rangeland results from the application of the sustainable
management strategy. For WA, by contrast, the expansion is connected to the high increase
in livestock numbers (Tab. 6.1), resulting from high population growth rates and assumptions on meat exports. When applying the sustainable management strategy for WA, which
was originally inferred from the storylines, the high increase in livestock numbers results
in LULCC scenarios, in which in 2050 all suitable area in Jordan and PA is covered by
70
6.6. Discussion and conclusions
rangeland (not shown). That is why we also applied the intensive management strategy
for this scenario. Even then, the results for the year 2050 indicate an unmet feed demand
for PA, because no suitable land is left on which rangeland could be allocated.
Figure 6.7 displays the land-use and land-cover map for the base year 2000 as well as
the land-use and land-cover maps for the different scenarios in 2050. For visualization
purposes, we aggregated the land-cover types fruits, vegetables, cereals/field crops, other
crops, and fallows to the category arable land, and the land-cover categories forests, natural
vegetation mosaic/croplands, scrublands, grasslands, woody savannahs, and wetlands to
the category (semi-)natural vegetation.
6.6. Discussion and conclusions
In this simulation experiment, we applied the regional-scale LandSHIFT version in order to
evaluate the effects of different assumptions on socio-economic developments in the Jordan
River region on land-use and land-cover change. We chose the integrated modeling system
LandSHIFT, because it was specifically designed to develop mid- to long-term, spatially
explicit LULCC scenarios. Furthermore, the modularized structure of LandSHIFT allowed
the easy exchange of subcomponents of the modeling system as well as the adjustment of
the represented processes of human decision making to different spatial resolutions. This
facilitated a straightforward integration of available land-use related information.
One aim of this simulation experiment was to develop LULCC scenarios in a spatial
resolution suitable to capture the variability of biophysical factors in the study region, e.g.
topography or the climatic gradient from north to south, and hence make the LULCC scenarios suitable to serve as input for subsequent impact assessments. The spatial resolution
of 1 km is a trade-off between the required level of detail, spatial coverage, data availability,
and computation time.
All four LULCC scenarios indicate an expansion of land-use area, inevitably associated
with a decrease in natural vegetation as well as alterations of the hydrological conditions
in the Jordan River region. Moreover, the scenarios MH and WA show high increases in
rangeland area, with MH assuming sustainable rangeland management connected with low
stocking densities and WA assuming intensive rangeland management resulting in high
stocking densities. Besides changes in leaf area index, evapotranspiration, and albedo, the
expansion of land-use activities may cause additional soil compaction and degradation of
land resources. For quantitative estimates of the environmental impacts, the simulation
results may be used as input for hydrological or environmental models. Nevertheless, a
general trend can be derived from the results of this simulation experiment: even though it
is the most optimistic scenario for the people in the Jordan River region, the WA scenario is
the one with the highest pressure on water resources and natural vegetation in the region.
A limitation of the presented LULCC scenarios is the difference in the spatial resolution
of the underlying crop yield information and the LULCC scenarios. While the scenarios were
calculated on a 1 km grid, the yield information was provided on a 30 arc minutes grid.
71
6. Future land-use and land-cover change scenarios for the Jordan River region
Fig. 6.7.: Land-use and land-cover maps for (a) the base year 2000, the year 2050 for the
(b) Modest Hopes (MH) scenario, (c) the Poverty and Peace (PP) scenario,
(d) the Suffering of the Weak and the Environment (SWE) scenario and (e) the
Willingness and Ability (WA) scenario.
72
6.6. Discussion and conclusions
Crop yield information with a higher spatial resolution would represent the spatial variability
of crop yields in the Jordan River region in a better way and consequently increase the
reliability of the LULCC scenarios. Moreover, the VALUE calculations base upon different
assumptions on crop yields, resulting in an unmet crop demand for the Bet She’an Basin in
the LULCC scenario calculations. This inconsistency can be solved by a comprehensive yield
raster for the entire study region, that is in an appropriate spatial resolution, preferably the
same as the LULCC scenarios, and suitable to serve as basis for the VALUE as well as the
LandSHIFT.R simulations. A limitation in the scenario assumptions, revealed by the unmet
feed demand for PA, is the high growth of livestock numbers caused by the assumed high
population growth rate for the WA scenario together with the assumption on meat export of
the IMPACT model. Even if rangeland is allocated with intensive rangeland management,
the livestock number is too high to be supported on the available area. However, sustainable
rangeland management must be considered an important issue in the Middle East, since
poorly managed production in the past has led to overgrazing and degradation in the
region (e.g. Abahussain et al. (2002); Al-Jaloudy (2001)). This is typically accompanied
by a reduction of infiltration capacity and an increase in surface runoff. An additional
iteration step in the scenario analysis would provide the opportunity to make estimates on
livestock numbers more plausible. This could be realized by adjusting the parameterization
of LandSHIFT.R, e.g. the fraction of crop residues in sheep and goat nutrition or additive
feed from other sources, or by an adaptation of the scenario assumptions, e.g. by adapting
import and export rates for meat. Moreover, the scenarios can be enriched by an inclusion
of climate change effects in addition to the socio-economic impacts on the landscape.
To sum up, the LULCC scenarios show how different assumptions on socio-economy
might affect land use and the distribution of natural vegetation in the Jordan River region.
Thus they supply decision makers with information to prepare future developments and/or
take countermeasures to mitigate unwanted developments. The LandSHIFT.R output in
form of maps and indicators is suited to visualize the impact of specific scenario assumptions on the landscape in the context of the scenario analysis. Hereby, the results of this
simulation experiment support the adjustment of the storylines and key factor quantifications. In combination with the detailed raster maps on relative human appropriation of
net primary production and population density, these scenarios can serve as an excellent
basis for the spatially explicit assessment of ecosystem services and their incorporation in
the development of pathways to increase the benefits from regional water resources for
both humans and ecosystems. Furthermore, the spatial resolution qualified the scenarios
to serve as basis for hydrological modeling and other environmental impact studies. Hence,
the LULCC scenarios show how the integrated modeling system LandSHIFT can be used
to provide input to the planning of sustainable water management strategies in a region
suffering from severe water stress.
73
7. Assessment of future conflicts
between agricultural land use
and biodiversity in Africa
This chapter describes the estimation of the area potentials for crop cultivation and livestock grazing in Africa. Moreover, an analysis of the possible threat to future agricultural
development to biodiversity and the effects of biodiversity conservation strategies on landuse and land-cover change is delineated. For this simulation experiment, we applied the
global-scale LandSHIFT version. The results indicate that there is a land reserve of more
than 2 000 million hectare for crop cultivation under rain-fed conditions, about 2 800 million hectare for crop cultivation under irrigated conditions, and approximately 1 250 million
hectare for livestock grazing. For all three activities, we identified a significant overlap of
suitable agricultural area with regions of high vascular plant diversity. This indicates that it
is indeed important to develop large-scale management and/or conservation strategies to
protect the species diversity of an ecosystem in order to maintain full ecosystem functions
and resilience.
7.1. Motivation
According to recent statistical and remote-sensing based surveys, about 8 % of the land area
of the African continent is used as cropland and another 30 % as rangeland (Ramankutty
et al., 2008). Still, agricultural production in some sub-Saharan regions is not sufficient
to fulfill human food demands, resulting in high rates of malnutrition (Rosegrant et al.,
2005a). The improvement of availability and access to food in these regions will be a crucial
issue in the coming decades. Pictures of possible future development trends were drawn in
a variety of scenario exercises such as the Millennium Ecosystem Assessment (Millennium
Ecosystem Assessment, 2005b) or the Global Environmental Outlook (Rothman et al.,
2007). Most of these scenarios indicate a strongly increasing human population in the
majority of African countries over the next decades.
In order to meet the growing demand for food, the agricultural area is likely to further
expand. The rate and location of cropland expansion will depend on technological change
to increase crop yields as well as on possible effects of a changing climate conditions,
whereas major drivers of rangeland expansion include livestock management strategies and
human diet preferences (Msangi and Rosegrant, 2009). Land-use and land-cover changes
75
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
are acknowledged as important negative influences on biodiversity and thus on the ability
of ecosystems to provide crucial goods and services (Foley et al., 2005; Metzger et al.,
2006; Biggs et al., 2008).
In order to analyze future conflicts between agricultural activities and biodiversity conservation, it is essential to identify the location of suitable land for agriculture (land reserve)
and to relate these locations to the spatial distribution of regions with a high degree of
biodiversity. There have been several attempts to assess the potentials for agriculture.
For example, Ramankutty et al. (2002) developed a model to quantify the potentially
cultivable land at the global scale by defining environmental envelopes that consider climatic constraints and soil constraints for crop cultivation. Other studies, such as WBGU
(2008), Fischer et al. (2008), and Hoogwijk et al. (2005), concentrate on area potentials for biofuels. For these studies, technically more sophisticated process-based modeling
approaches were applied. Additionally, they impose different types of constraints for the
use of conservation areas as cropland. Nevertheless, they do not explicitly address spatial
correlations between land suitable for agricultural activities and regions with a high degree
of biodiversity.
The objectives of our simulation experiment were twofold: First, it gives an improved
estimate of potential agricultural area in Africa and correlates this area to the spatial pattern
of vascular plant diversity under current climate conditions. In contrast to the abovementioned assessments, which concentrate on cropland, suitabilities for both, rangeland
and cropland, are considered. Secondly, an exemplary trajectory of land-use and land-cover
change and its effects on habitat loss is analyzed. Based on this, the effectiveness of existing
conservation areas is tested. Moreover, it is analyzed to what extent the inclusion of new
conservation areas can help to reduce habitat loss and what this would imply for location
and quantity of land-use and land-cover change. The simulation experiment consists of
a set of simulation runs for the whole African continent conducted with the global-scale
version of the integrated modeling system LandSHIFT. The simulation experiment aims to
identify ways to minimize conflicts between agricultural development and conservation of
biodiversity.
7.2. Materials and Methods
7.2.1. Input data and model initialization
The initial land-use and land-cover map used by LandSHIFT is based on the global IGBP
land-cover dataset, derived from AVHRR source imagery data (Loveland et al., 2000). The
content of this map was extended by additional information on the spatial distribution
of crop types generated with a procedure that merges land-cover data with sub-national
census data (Heistermann, 2006).
The LandSHIFT routines for preference ranking and land allocation use micro-level information on landscape characteristics, zoning regulation, and land-use related model va-
76
7.2. Materials and Methods
riables. Slope information bases upon on the HYDRO1k data set (USGS, 1998), while the
river network density was calculated as line density of streams per grid cell based on the
HYDRO1k data set (USGS, 1998). In order to derive information on zoning regulation,
we mapped the micro-level grid cells to data sets on areas designated as national or international conservation areas (WDPA Consortium, 2004). Furthermore, spatial information
on Tsetse fly distribution was considered in order to constrain livestock grazing (Wint and
Rogers, 2000). Information on population density was derived from the History Database
of the Global Environment HYDE (Klein Goldewijk, 2005). Information on infrastructure
was obtained from the VMAP0 data set on roads (NIMA, 1997). Additionally, climate data
for the reference period (1971-2000) was taken from the “CRU TS 2.1” gridded dataset for
monthly precipitation, air temperature, cloud cover, and frequency of wet days (Mitchell
and Jones, 2005). The climate information was used by LPJmL (Sitch et al., 2003; Bondeau et al., 2007) in order to calculate crop yields under rain-fed and irrigated conditions
as well as NPP of rangeland and natural vegetation. For the period of investigation, the
climate input parameters were kept constant. The resulting maps with a spatial resolution
of 30 arc minutes were geographically mapped to the micro-level grid cells of the globalscale LandSHIFT version. Additionally, these maps comprise information on the share of
irrigated area for each crop type derived from the FAO statistical database (FAO, 2009).
7.2.2. Geographic distribution of vascular plant diversity
In this simulation experiment, we used information on patterns of vascular plant diversity
for Africa, which was derived from the global map on species numbers of vascular plants
(Barthlott et al., 2007; Mutke and Barthlott, 2005). The map was generated using an
inventory approach based on summary data on the floras of operational units (Mutke and
Barthlott, 2005). The mapping builds on the species-area model of Arrhenius (1920, 1921).
Extinctions or invasions were not taken into account. At the global scale, the spatial distribution of vascular plants is strongly correlated with the distribution of vertebrate richness
and insect diversity. Therefore, we use vascular plant diversity in a first approximation as
indicator for overall biodiversity.
For the African continent, the map of vascular plant diversity depicts nine diversity zones
(DZ), with DZ one representing the lowest species diversity (< 100 species per 10 000
km2 ) and DZ nine representing the highest species diversity (4 000 - 5 000 species per
10 000 km2 ). In Africa, the occurrence of vascular plant diversity follows the latitudinal
gradient, with centers of vascular plant diversity in the Albertine Rift, Cameroon and
Guinea, Madagascar, and the coastal regions of southern Africa.
In this simulation experiment, we show exemplarily how spatial information on vascular
plant diversity can be used in order to develop large-scale conservation strategies. For this
purpose, we translated the diversity patterns into a land-use constraint in LandSHIFT. As
described in equation 3.2 (see chapter 3), a constraint affects the suitability value of a
micro-level grid cell. The higher the DZ number and, thus, the species number of vascular
plants per grid cell, the higher is the constraint’s impact on the suitability. As a result,
77
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
areas located in high DZs are less suitable for land-use activities and subject to higher
conservation. Additionally, it was assumed that grid cells formerly used as rangeland are
more suitable for METRO and AGRO than grid cells with natural vegetation.
7.2.3. Estimation of area potentials for agriculture and their
spatial correlation to regions with high vascular plant
diversity
In this simulation experiment, we analyze the area potentials for three agricultural activities
in Africa: crop production under rain-fed conditions (AGRO RF), crop production under
irrigated conditions (AGRO IR), and livestock grazing (GRAZE). For this purpose, LandSHIFT calculates the amount of area potentially suitable for these activities, using the
first part of equation 3.2 (see chapter 3), under consideration of the suitability factors and
weights listed in Tab. 7.1. The suitability assessment operates on country level. Hence, the
same combination of biophysical and socio-economic factors can lead to different suitability
values in different countries. In order to assess the area potentials for crop production, all
cells that are not classified as urban or cropland under baseline conditions were analyzed;
this includes cells classified as rangeland. The suitability maps were created in two steps.
First, suitability was calculated for all modeled crop types individually. Second, for the crop
types with rain-fed and irrigated yields exceeding a threshold value of 100 kilograms per
hectare, the medium value of their suitability was computed and displayed in a suitability
map. In case the yield of any crop type fell below the threshold value, its suitability was set
to zero. In order to assess the area potentials for livestock grazing, all cells were analyzed
that are not classified as urban, cropland or rangeland under baseline conditions, and where,
similar to the cropland analysis, the NPP exceeds a threshold value of 100 kilograms per
hectare.
Tab. 7.1.: Suitability factors and weights for the assessment of area potentials for agricultural activities in Africa.
Land-use activity
METRO
AGRO
GRAZE
78
Suitability factor
Slope
Infrastructure
Yield
Slope
Population density
Proximity to cropland
Net primary productivity
Slope
Population density
Proximity to cropland
River network density
Weight
0.4
0.6
0.25
0.25
0.25
0.25
0.2
0.2
0.2
0.2
0.2
7.2. Materials and Methods
The suitability maps for the three agricultural activities were geographically mapped to
spatial information on vascular plant biodiversity, using the ArcGIS software package. For
an easier presentation of the results, we divided the suitability values for these areas into
the four suitability categories highly suitable (suitability value > 0.75 - 1.0), suitable (> 0.5
- 0.75), moderately suitable (> 0.25 - 0.5), and marginally suitable (≥ 0 - 0.25). Finally
the distribution of the suitable micro-level grid cells over the different DZs was calculated.
7.2.4. Assessment of land-use and land-cover change impacts
on habitat loss
In order to investigate the impact of possible future land-use and land-cover changes on
biodiversity, we calculated the extent and location of land use under the UNEP Global
Environmental Outlook 4 (GEO4) scenario Sustainability First (Rothman et al., 2007).
This was done without any land-use constraint except for the transition constraint, which
is a proxy for pattern stability. In the following, we refer to this simulation run as SR one.
On the basis of the Sustainability First scenario, we made a set of second order assumptions, in order to investigate the impact of different conservation strategies on the extent
and location of land-use and land-cover change. Therefore, we combined the Sustainability
First scenario with land-use constraints for conservation areas and DZs. Altogether, we
performed three additional simulation runs with different second order assumptions: (1)
one run with a land-use constraint for the DZs (SR two), (2) one simulation run with a
strict exclusion of conservation areas (SR three), and (3) one run with a combination of a
constraint for the DZs and a strict exclusion of conservation areas (SR four).
7.2.5. Scenario drivers
Scenarios are plausible descriptions of how the future may unfold (Alcamo, 2008). The
scientific community applies scenarios in order to estimate and assess possible future states
of the environment or to examine the effect of alternative policy options. We derived
estimates on the future development of the driving forces of land-use and land-cover change
in Africa from the Sustainability First scenario of the GEO4 report (Rothman et al., 2007).
The base year of the scenario calculations is the year 2000. The GEO4 scenarios give
an idea of how current social, economic, and environmental trends might develop in the
future and vary regarding assumptions made on policy approaches and societal choices.
Sustainability First puts a much stronger emphasis on environmental and social concerns
and applies a more decentralized model of economic development as compared to the other
GEO4 scenarios. It assumes a high level of cooperation between civic, government, and
private sectors and strongly emphasizes equity.
Driving forces were derived from the Sustainability First scenario and comprise changes
in human population, crop production, and livestock numbers. Population scenarios were
computed with the IFs model (Hughes, 1999). Accordingly, Africa’s population increases
79
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
from approximately 0.8 billion in 2000 to about 1.16 billion in 2020. The future agricultural
demand was computed with the IMPACT model (Rosegrant et al., 2008). Production of
the major crop types (wheat, maize, other grains, soybeans, millet, and sorghum) increases
from about 77 million metric tonnes to 135 million metric tonnes. Likewise, the production
of grazing livestock increases from about 70 million livestock units in 2000 to 100 million
livestock units up to 2020. The Sustainability First scenario assumes better nutrition and
diminishing of malnourishment; the calorie availability per capita and day is assumed to
increase from below 2 000 up to about 3 000. In order to single out the potential effect
of agricultural expansion, yield increase due to technological change was not taken into
account. For a detailed description of the GEO4 scenarios, see Rothman et al. (2007).
We chose the Sustainability First scenario, because the storyline of this scenario with its
strong focus on sustainable resource use, maintenance of ecosystem services, and species
conservation is consistent with the objective of this study: The conservation strategies
tested in this simulation experiment are most likely to be realized under a future as described
in this scenario.
7.3. Results
7.3.1. Area potentials for agricultural activities
Figure 7.1 displays the baseline suitability maps for the activities crop cultivation under
rain-fed (AGRO RF) as well as irrigated conditions (AGRO IR), and livestock grazing
(GRAZE). Under baseline conditions (Fig. 7.2) approximately 2 064 million hectares are
available, potentially suitable for crop production under rain-fed conditions. Table 7.2 lists
the partitioning of this area into the four suitability categories described in section 7.2.3.
Figure 7.3 displays the distribution of the land reserve for crop cultivation under rainfed conditions over the DZs. With about 1.3 % in DZ eight and 0.4 % in DZ nine,
only a small fraction of the area potentially suitable for crop cultivation under rain-fed
conditions is located in DZs with the highest vascular plant diversity. The distribution of
the calculated average suitability values for the areas potentially suitable for crop cultivation
and the corresponding standard deviations over the DZs is shown in Fig. 7.4. In general, the
suitability values for crop cultivation under rain-fed conditions are higher than the values
for crop cultivation under irrigated conditions, except for the values in DZ one. Figure 7.5A
shows the division of the area potentially suitable for AGRO RF, located in the nine DZ,
into the four suitability categories. Obviously, in all DZs, the largest fraction of the areas is
classified as marginally suitable and moderately suitable. No part of the area located in DZ
one is categorized as highly suitable or suitable. Diversity zone nine features the largest
fraction of suitable or highly suitable area. Except for DZ three, there is a continuous
increase in fraction of area categorized as suitable from DZ two (5.6 %) to DZ nine (31
%).
Under baseline conditions, about 2 784 million hectares potentially suitable for crop
80
7.3. Results
Fig. 7.1.: Suitability maps under baseline conditions for the agricultural activities crop production under rain-fed conditions (AGRO RF), crop production under irrigated
conditions (AGRO IR), and livestock grazing (GRAZE).
Tab. 7.2.: Division of the area potentials under baseline conditions for crop cultivation
under rain-fed conditions (AGRO RF), crop cultivation under irrigated conditions
(AGRO IR), and livestock grazing (GRAZE) into the four suitability categories.
Suitability category
Highly suitable
Suitable
Moderately suitable
Marginally suitable
Land-use activity
AGRO RF AGRO IR GRAZE
Area
Area
Area
(mio ha)
(mio ha) (mio ha)
19.94
18.36
0.02
242.58
238.10
99.79
825.83
927.97
1 104.06
975.56
1 599.25
41.20
Land-use activity
AGRO RF AGRO IR GRAZE
Fraction
Fraction Fraction
(%)
(%)
(%)
1.0
0.7
0.0
11.8
8.6
8.0
40.0
33.3
88.7
47.3
57.5
3.3
production under irrigated conditions are available (Tab. 7.2). Approximately 87 % of this
area is located in the DZs two to six, only 1 % is located in DZ eight and 0.3 % is located
in DZ nine (Fig. 7.3). Figure 7.5B displays the division of the residual area potentially
suitable for AGRO IR, located in the nine DZs, into the four suitability categories. Again,
the largest part in the different DZs is categorized as marginally suitable or moderately
suitable. Diversity zone one includes no area categorized as suitable or highly suitable.
There is a continuous increase in suitable area from DZ two (1 %) to DZ nine (30 %).
With about 2 %, the largest fraction of highly suitable area is allocated to DZ nine.
81
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
Fig. 7.2.: Land-use and land-cover map for the base year (1993).
Out of 2 064 million hectares, potentially suitable for AGRO RF and 2 784 million hectares, potentially suitable for AGRO IR, 654 and 694 million hectares, respectively, are used
as rangeland under baseline conditions. The largest fraction of these areas (82 %) is located
in the DZs three to six (Fig. 7.6). About 85 % and 86 %, respectively, are categorized as
moderately suitable or marginally suitable. Only about 1 % is classified as highly suitable
(Tab. 7.3).
Under baseline conditions, 1 245 million hectares are potentially suitable for grazing.
The major part of this area is classified as moderately suitable (Tab. 7.2). Only a relatively
small area of 24 729 ha is categorized as highly suitable for livestock grazing. The results
for the distribution of the residual area potentially suitable as rangeland over the DZs
show similar trends as the results for crop cultivation (Fig. 7.3). Only a minor part of
the residual area potentially suitable for GRAZE is located in DZ eight (1.3 %) and DZ
nine (0.6 %). The average suitability of the areas potentially suitable for GRAZE shows in
general higher values and a lower variability as the corresponding values for crop cultivation
(Fig. 7.4). There is an increase in average suitability values over the DZs, except for the
DZs two, three and seven. Areas with the highest average suitability for grazing are located
82
7.3. Results
Fig. 7.3.: Distribution of area potentials over the diversity zones, for crop production under rain-fed conditions (AGRO RF), crop production under irrigated conditions
(AGRO IR), and livestock grazing (GRAZE). “–” labels areas that are not located
in any of the nine diversity zones.
Tab. 7.3.: Division of the area potentials, located on areas classified as rangeland under baseline conditions, for crop cultivation under rain-fed conditions (AGRO RF) and
crop cultivation under irrigated conditions (AGRO IR) into the four suitability
categories.
Suitability category
Highly suitable
Suitable
Moderately suitable
Marginally suitable
Land-use activity
AGRO RF AGRO IR
Area
Area
(mio ha)
(mio ha)
7.25
6.38
93.21
91.07
276.65
282.91
276.51
313.54
Land-use activity
AGRO RF AGRO IR
Fraction
Fraction
(%)
(%)
1.1
0.9
14.3
13.1
42.3
40.8
42.3
45.2
in DZ nine. Figure 7.5C shows the subdivision of the residual area potentially suitable as
rangeland, located in the nine DZs, into the suitability categories. Only small fractions of
the area are categorized as highly suitable or marginally suitable for grazing. No part of the
areas located in the DZs one and nine is classified as marginally suitable. Furthermore, no
83
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
Fig. 7.4.: Distribution of the average suitability values and the corresponding standard
deviations for crop production under rain-fed conditions (AGRO RF), crop production under irrigated conditions (AGRO IR), and livestock grazing (GRAZE)
over the diversity zones.
part of the area located in DZ one is categorized as suitable or highly suitable. In general,
the fraction of area classified as suitable, increase continuously from DZ two (0.1 %) to
DZ nine (36 %).
7.3.2. Scenario analysis
The Fig. C.1 through C.4 in appendix C display the land-use and land-cover distribution in
2020 for the four simulation runs (SRs) as well as the corresponding changes in land use
and land cover as compared to the baseline (1993). Under baseline conditions, urban and
built-up area covers about 10 mio ha, cropland demands 192.9 mio ha, rangeland covers
693.9 mio ha, leaving approximately 2 087 mio ha of natural vegetation (Tab. 7.4). The
results for the four SRs reveal an expansion for all land-use categories until 2020. In general,
the SRs one and three, and the SRs two and four show very similar simulation results. With
100 mio ha additional cropland for SR one up to 152 mio ha additional cropland for SR
four, the most obvious expansion occurs for the land-use activity AGRO.
An expansion of land-use activities is connected to a loss of natural vegetation: For the
SRs one and three, the loss of area covered with natural vegetation amounts to approx.
84
7.3. Results
Fig. 7.5.: Division of the area potentials, located in the different diversity zones, for crop
production under rain-fed conditions (A), crop production under irrigated conditions (B), and livestock grazing (C) into the four suitability categories.
85
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
Fig. 7.6.: Distribution of area potentials, located on areas classified as rangeland under
baseline conditions, for crop production under rain-fed conditions (AGRO RF),
crop production under irrigated conditions (AGOR IR), and livestock grazing
(GRAZE) over the diversity zones. “–” labels areas that are not located in any of
the nine diversity zones.
140 mio ha, and for the SRs two and four it amounts to approx. 180 mio ha. The largest
area expansion is calculated for the land-use activity AGRO, the smallest area expansion
is calculated for the land-use activity METRO. For the land-use activity GRAZE, the
additional area demand ranges from 24.1 mio ha for SR four to 36.1 mio ha for SR one
(Tab. 7.4). Altogether, for the SRs two and four 40 mio ha more additional area are required
than for the SRs one and three, however, for the SRs one and three 12 mio ha additional
area located in the DZs seven to nine are required (Tab. 7.4).
The breakdown of urban and built-up area in the different DZs is given in Fig. 7.7A.
All SRs in all DZs show an expansion of area as compared to the baseline. The strongest
increase in extent of urban and built-up area occurs in DZ four (ranging from 1.17 mio ha
for SR three to 1.5 mio ha for SR two), the weakest increase occurs in DZ one (ranging
from 50 ha for SR three to 145 ha for SR two). In the DZs one to five, the SRs two and
four, show a stronger increase in area demand as compared to SR one and three. This
is also true for DZ eight. In contrast, for all other DZs, the SRs one and three exhibit a
higher area demand than the SRs two and four. Figure 7.7B displays urban and built-up
area as fraction of the total area of the different DZ. The DZs seven and eight feature the
86
7.3. Results
Tab. 7.4.: Area statistics for METRO, AGRO, GRAZE, and natural vegetation under baseline conditions and for the four simulation runs (SR).
Land-use or
land-cover
category
METRO
AGRO
GRAZE
Natural vegetation
Simulation
run
Baseline
SR 1
SR 2
SR 3
SR 4
Baseline
SR 1
SR 2
SR 3
SR 4
Baseline
SR 1
SR 2
SR 3
SR 4
Baseline
SR 1
SR 2
SR 3
SR 4
Year
Area
1993
2020
2020
2020
2020
1993
2020
2020
2020
2020
1993
2020
2020
2020
2020
1993
2020
2020
2020
2020
(mio ha)
10.0
15.0
15.9
15.0
15.9
192.9
293.0
344.0
295.3
344.9
693.9
730.0
719.6
729.4
715.3
2 087.3
1 946.2
1 904.6
1 944.5
1 908.0
Change compared
to baseline
(%)
0.0
50.3
59.3
50.4
59.6
0.0
51.8
78.3
53.0
78.8
0.0
5.2
3.7
5.1
3.1
0.0
-6.8
-8.8
-6.8
-8.6
highest values of DZ area in use, with more than 1 %.
Figure 7.8A shows the area used for crop cultivation subdivided in the different DZ. In
the DZs eight and nine, all four SR feature a decrease in cropland as compared to the
baseline. This is also true for the SRs one and three in DZ one. For the remaining DZs, the
results indicate an increase in cropland extent. The strongest area increase is calculated for
DZ five, ranging from 42.68 mio ha for SR one to 58.1 mio ha for SR four. The weakest
increase as compared to the baseline is calculated for DZ nine, with 0.38 mio ha for the
SRs one and three. In the DZs one to five, the calculated area demand for the SRs two
and four clearly exceed the one of SR one and SR three. The reverse is true for the DZs
six to nine. Figure 7.8B illustrates the fraction per DZ used as cropland. The DZs eight
and nine feature the highest fraction of area used as cropland, whereas the least fraction
of cropland is located in DZ one and two.
The distribution of rangeland over the DZs is given in Fig. 7.9A. Obviously, the largest
part of rangeland is located in the DZ three, four and five (ranging from altogether 587 mio
ha for SR four to 608 for SR one), while only a marginal part of the rangeland is located
in DZ one (0.01 mio ha for SR one and three and 0.25 mio ha for SR two and four) and
87
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
Fig. 7.7.: (A) Distribution of the area demand calculated for the land-use activity METRO
over the diversity zones, for baseline conditions as well as for the four simulation
runs in the year 2020 and (B) share of these areas in the total area of the different
diversity zones.
88
7.3. Results
Fig. 7.8.: (A) Distribution of the area demand calculated for the land-use activity AGRO
over the diversity zones, for baseline conditions as well as for the four simulation
runs in the year 2020 and (B) share of these areas in the total area of the different
diversity zones.
89
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
DZ nine (1.51 mio ha for SR one and three and 1.57 mio ha for SR two and four). For
the DZs seven to nine, the simulation results indicate only a small increase in rangeland
as compared to the baseline, for the DZs five and six, the results even indicate a decrease
in rangeland as compared to the baseline. For the DZs two, six and nine, the rangeland
extent for the SRs two and four exceeds the extent calculated for the SRs one and three.
For the other DZ the opposite is true. Figure 7.9B shows the fraction in the different DZs
used as rangeland. As for the other land-use activities, a small fraction of the available
area in the DZs one and two is used as rangeland. In contrast to the land-use activities
METRO and AGRO, not the high DZ but the DZ three and four show the highest fraction
of use as rangeland, since GRAZE is the most area intensive land-use activity of the three
simulated land-use activities.
A detailed analysis of the areal changes in DZ nine reveals, that for the SRs two and four,
both applying a restricted use of high DZs, no additional area is converted as compared
to the baseline. Increases in urban and built-up area (0.94 mio ha) and in rangeland (1.41
mio ha) in this DZ are compensated by a decrease in cropland (2.35 mio ha).
7.4. Discussion
In this simulation experiment, we showed how a dynamic, spatially explicit land-use change
model can be applied in order to quantify the area potentials for crop cultivation and
livestock grazing in Africa. Moreover, we analyzed the possible threat of future agricultural
development to biodiversity and tested the effect of biodiversity conservation strategies on
land-use and land-cover change. This kind of simulation experiment is comparable to those
of Rouget et al. (2003), who assessed the area losses for different biome types caused by
land-use change and species invasion, for a region in South Africa.
A central element of our analysis is the calculation of crop yields and NPP of rangeland and natural vegetation. For this task, we applied the process-based vegetation model
LPJmL, which proved its capacities in a comparable study, related to the assessment of
bioenergy potentials (WBGU, 2008). We think that the process-based approach is a major
step forward compared with the environmental envelopes approach, implemented by Ramankutty et al. (2002), since the former captures biophysical processes of plants in more
detail and enables the consideration of a variety of crop types.
The definition of suitability, applied in this simulation experiment, incorporates environmental as well as socio-economic factors, which were recognized as fundamental drivers
of land-use change (Geist and Lambin, 2001). The suitability values were calculated on
country level, which implies that the same combination of biophysical and socio-economic
factors can lead to different suitability values in different countries. Since LandSHIFT performs the spatial allocation of agricultural activities on the country-level, this simulation
experiment represents a regionally more diverse view than assessments calculating absolute numbers for the whole continent (Ramankutty et al., 2008; WBGU, 2008). To our
knowledge, our analysis is the first of its kind, which explicitly considers crop cultivation
90
7.4. Discussion
Fig. 7.9.: (A) Distribution of the area demand calculated for the land-use activity GRAZE
over the diversity zones, for baseline conditions as well as for the four simulation
runs in the year 2020 and (B) share of these areas in the total area of the different
diversity zones.
91
7. Assessment of future conflicts between agricultural land use and biodiversity in Africa
under rain-fed and irrigation conditions as well as livestock grazing as separate activities
and, thus, gives a comprehensive overview of the agricultural sector.
In our simulation experiment we found, that under current climate conditions there is a
land reserve of more than 2 000 million hectare which is suitable for rain-fed crop cultivation. This is more than four times the area identified by Ramankutty et al. (2002). This
difference is reduced if we leave out areas categorized as marginally suitable. Nevertheless,
the resulting area is still more than twice as much as the reference value. One reason might
be that we used a process-based vegetation model for yield estimates and differentiated
between crop types instead of using rather aggregated environmental envelopes, but left
out soil constraints. Another reason is the threshold value of 100 kilograms per hectare
for crop yields as well as for NPP of natural vegetation and rangeland. The area potential
for irrigated crop cultivation is about 2 800 million hectare. It has to be noted that these
value does not say anything about the feasibility of irrigation, since water availability and
competition to other water use sectors were not taken into account (Alcamo et al., 2003).
Looking at the area potential for livestock grazing, we found a large overlap with potential
cropland area. This indicates a possible competition between these agricultural activities
for the same land resources. Note, for livestock grazing the assessment of potential area
did not consider constraints due to Tsetse fly occurrence, which plays an important role for
cattle in Africa (Cecchi et al., 2008). In the land-use change simulations, we have taken
this effect into account via a land-use constraint for affected areas.
We could identify an increase in suitability for agricultural activities (both crop cultivation and livestock grazing) in correlation with increasing vascular plant diversity. This
indicates that areas with a high degree of biodiversity are threatened by agricultural expansion and urbanization, supposing that most suitable areas are used preferably. This
threat was revealed by the simulation experiments one and two, which show a major habitat loss in the DZs ≥ 5, provided that land-use activities are not restricted in these DZs.
We furthermore revealed indirect land-use and land-cover change. Regions formerly used
for livestock grazing were converted to cropland, and the rangeland was shifted to areas
formerly covered with natural vegetation.
Another striking result of our simulation experiment is that the inclusion of a biodiversity
constraint in addition to existing conservation areas did not results in a high increase of land
conversion (plus 1.5 % of the total study area), but to a shift within the spatial patterns of
land-use. This might lead to the conclusion that the creation of new protection areas will
not substantially hinder the development for agriculture, at least not from the continental
perspective. Here, the protection of high DZs from agricultural and urban expansion is a
relatively simple assumption, as it does not consider regional characteristics as for example
endemic species, but in its core it can be related to the approach of Küper et al. (2004),
who also define biodiversity hot spots as potential protection targets.
A major simplification of our experimental design is that we did not consider climate
change, which is likely to have a strong impact on both, crop yields (Parry et al., 2003)
and spatial patterns of species distribution (Harrison et al., 2006), and rather focused on
the effects of changes in socio-economic drivers of land-use change. For this reason we
92
7.5. Conclusions
limited the time horizon of our simulation experiment to 2020. For long-term analysis, we
think that both effects need to be taken into account as recent simulation studies show
major changes of climate parameters after 2050 across scenarios (IPCC, 2007). A long-term
analysis would furthermore require a far more sophisticated assessment approach, in order
to account for the overlapping dynamics of land-use change and ecosystem adaptation
processes to changing climate conditions. Another simplification is, that our model drivers
neglect technological change in order to single out the potential effect of agricultural
expansion. This idealized assumption was made in order to emphasize the overall dynamics.
Another possible improvement of our study design is motivated by publications showing
that sustainable agricultural management has the potential to enhance biodiversity and
ecosystem functions (e.g. Tscharntke et al. (2005)). Here, we could use our model to
analyze trade-offs between intensive agriculture (and possible stronger negative effects on
biodiversity) and the implementation of low intensity agricultural systems, accompanied by
less negative impacts but a larger area demand. A major prerequisite for that type of study
is the use of more sophisticated indicators to measure the human influence on biodiversity.
Currently our study is limited to the simplistic assumption that conversion to cropland or
rangeland leads to the loss of natural vegetation. In this context, the indicators developed
for example by Scholes and Biggs (2005) and Luck et al. (2004) that establish a relationship
between intensity of land use and its impacts on biodiversity as well as the human-biomes
approach of Ellis and Ramankutty (2008) will be good starting points for further research.
As LandSHIFT already calculates cell level densities of human population, crop production,
and stocking density, these concepts can be implemented within the existing modeling
philosophy.
7.5. Conclusions
In our simulation experiment, we could identify a significant overlap of suitable agricultural
area and vascular plant diversity. This indicates that it is indeed important to develop
large-scale management and/or conservation strategies to protect the species diversity of
an ecosystem in order to maintain full ecosystem functioning and resilience. These efforts
have to go hand in hand with the development of sustainable agricultural management
and the principal aim to produce enough food for human population. Our experiments
show that in the medium term there is the potential to achieve both goals. In the longterm perspective population pressure becomes even larger. Together with changing climate
conditions, this will very likely lead to large impact on both, natural ecosystems and agroecosystems.
93
8. Synthesis
The overall objective of this thesis was to investigate management strategies for Africa and
the Jordan River region that allow a resource-conserving production of agricultural commodities. For this purpose, a new regional-scale version of LandSHIFT was constructed
and the global-scale LandSHIFT version was refined. In a set of simulation experiments,
the two LandSHIFT versions were applied in order to test the new functionalities and to
investigate different strategies for resource-conserving agriculture and nature conservation.
In this chapter, additionally to the thematic discussions provided in chapters 4 to 7, the findings of the different simulation experiments are summarized and research needs emerging
from the integrated consideration of these simulation experiments are highlighted.
8.1. Summary of findings
There is a growing awareness that the ability of the Earth’s ecosystems to provide goods
and services is largely dependent upon the well-functioning of these systems. It has been
shown that expansion and intensification of agricultural activities may adversely affect
ecosystems through disruption or alterations in the composition of the systems. This may
result in a serious threat of future food production and human well-being. Hence, there
is an urgent need for ecologically based management strategies that help to meet human
demands without compromising environmental integrity. The overall objective of this thesis
was to investigate such ecologically based management strategies, focusing on rangeland
management in the Jordan River region and nature conservation strategies in Africa.
In order to investigate the effects of different rangeland management strategies on the
ecosystems of the Jordan River region, a regional-scale LandSHIFT version was constructed. This LandSHIFT version is able to simulate the feedback between stocking density
and biomass productivity and the restriction of stocking densities based on stocking capacities. The new functionalities of LandSHIFT.R were tested (chapter 4) and methods to
quantify the environmental impact of different grazing intensities were analyzed (chapter
5). Furthermore, the revised LandSHIFT version was validated and applied in the context
of an environmental scenario analysis in order to investigate potential future development
pathways for the Jordan River region (chapter 6).
In order to analyze the reconcilability of food production and nature conservation in
Africa, the global-scale LandSHIFT version was extended by an algorithm that allows to
gradually constrain land-use activities in regions with high degrees of biodiversity. The
revised global-scale LandSHIFT version was then applied to evaluate the area potentials
95
8. Synthesis
for agricultural activities in Africa and to investigate a set of continental-scale nature
conservation strategies (chapter 7).
In this final chapter, the most relevant conclusions resulting from previous chapters are
reflected, main achievements and limitations are highlighted, and research needs emerging
from the integrated consideration of these simulation experiments are pointed out.
8.1.1. Modeling the feedback between stocking density and
biomass productivity
In an initial simulation experiment, LandSHIFT.R was applied for examining the impact
of different rangeland management strategies on the dynamics of grazing systems in the
Jordan River region. Based on various assumptions on rangeland management strategies
and on landscape degradation due to overgrazing, the possible future development of landuse and land-cover change was assessed.
The results of this simulation experiment proved that the computed land-use and landcover changes are sensitive to the implemented feedback mechanism and to the different
rangeland management strategies. However, the validation of the calculated rangeland
extent revealed that LandSHIFT.R underestimates the area demand for rangeland. One
reason for that mismatch might be that grazing of goats and sheep in the Middle East
often takes place on (semi-)natural vegetation and, hence, the statistical data for permanent meadows and pastures (FAOSTAT) can only be a rough estimate for rangeland
area. Furthermore, the validation of stocking density distribution was hampered, because
no spatial data on stocking densities in an appropriate spatial resolution was available.
Consequently, it must be concluded that LandSHIFT.R is in principle capable of simulating
different rangeland management strategies and the feedback between stocking density and
biomass productivity, but that their quantity is subject to a wide range of uncertainty.
Further model improvement will be achieved by a refinement of the temporal resolution,
which gains more importance due to the implemented feedback mechanism, and the consideration of soil processes that play an important role for degradation processes (Ibáñez
et al., 2007), but are currently not considered in the model.
8.1.2. Quantifying the environmental impact of grazing in
Jordan
In this simulation experiment, LandSHIFT.R was used to carry out two simulation runs
for Jordan, differing only regarding the applied rangeland management strategy. These
simulation runs were evaluated with a set of landscape indicators and an indicator for
pressures on biodiversity in order to quantify the impact of different grazing intensities on
biomass productivity and landscape structure.
The results of this simulation experiment showed that the rangeland area demand resulting from the application of sustainable rangeland management exceeds the one resulting
96
8.1. Summary of findings
from the application of intensive rangeland management by far. However, the evaluation
of the corresponding distribution of stocking densities with the indicator for pressures on
biodiversity, relative HANPP, revealed that the application of intensive rangeland management resulted in a higher average value of relative HANPP (58 %) as compared to the
application of sustainable rangeland management (47 %). Moreover, the evaluation of the
rangeland area with a combination of relative HANPP and landscape pattern metrics implied a stronger fragmentation of the landscape for the simulation run applying an intensive
rangeland management. One limitation of the combined analysis of landscape fragmentation is the choice of a threshold value for relative HANPP. An improvement strategy could
be to derive the threshold value as a function of climate conditions (e.g. coefficient of
variation for inter-annual precipitation).
However, the applied indicators proved suitable to reveal secondary effects resulting from
high stocking densities, such as vegetation degradation, which are not apparent from landuse and land-cover maps. Furthermore, the combination of relative HANPP and landscape
pattern metrics can support the development of moderate rangeland management strategies in order to maintain open landscapes in the Mediterranean region, without causing
adverse effects such as the fragmentation of native habitats.
8.1.3. Future land-use and land-cover change scenarios for the
Jordan River region
In this simulation experiment, the revised regional-scale LandSHIFT version was validated
and then applied in the context of an environmental scenario analysis. The objective of the
scenario analysis was to investigate potential future development pathways for the Jordan
River region. Altogether, four simulation runs were conducted, varying in socio-economic
drivers and rangeland management strategy.
The validation of the enhanced LandSHIFT.R version indicates that the model preferentially simulates changes in land-use and land-cover at locations where land-use and
land-cover changes were actually observed. Hence, it can be concluded that LandSHIFT.R
provides a reliable assessment of land-use changes.
In all four simulation runs an expansion of agricultural activities took place. The simulation run associated with the scenario Willingness and Ability showed the strongest expansion
of agricultural area even though intensive rangeland management was applied. Hence, the
Willingness and Ability is the most optimistic scenario for the people in the Jordan River
region; but, at the same time, it is the one with the highest pressure on natural resources.
One limitation of the presented scenario calculations is the low spatial detail of the
underlying crop yield information. Crop yield information in a higher resolution would
better represent the spatial variability of biophysical factors and consequently increase the
reliability of the simulation results.
However, the regional-scale LandSHIFT version is the first land-use change model that
covers the entire Jordan River region and generates simulation results in a spatial reso-
97
8. Synthesis
lution suitable to serve as a basis for environmental impact studies. The inclusion of the
feedback mechanism between stocking density and biomass productivity and the evaluation
of stocking densities in form of raster maps on relative HANPP makes these innovative
simulation results suitable to serve as basis for the spatially explicit analysis of ecosystem
goods and services. The consideration of these simulation results for the development of
regional agricultural management strategies can help to increase the benefits from natural
resources for both, humans and ecosystems, and to maintain ecosystem functions in the
long run.
8.1.4. Assessment of future conflicts between agricultural land
use and biodiversity in Africa
In this simulation experiment, the revised global-scale LandSHIFT version was applied in
order to quantify the area potentials for crop production and livestock grazing in Africa. Moreover, the possible threat of future agricultural development to biodiversity was
analyzed.
The simulation results revealed that under current climate conditions there is an area
potential of about 2 000 million hectare for rain-fed crop production. This is more than four
times the area identified by Ramankutty et al. (2002). Possible causes for the mismatch
are the application of a process-based vegetation model for yield estimates instead of
environmental envelopes and the threshold value of 100 kilograms per hectare for crop
yields and NPP of natural vegetation and rangeland considered in the simulation runs. The
inclusion of a biodiversity constraint did not result in a strong increase of land conversion.
Hence, from the continental perspective, the creation of new protection areas is not very
likely to substantially hinder the development for agriculture.
One limitation of the experimental design is that changing climate conditions were neglected. However, climate change is likely to have a strong impact on crop yields and spatial
patterns of species distribution. Hence, the inclusion of changes in climate parameters in
the context of a long-term analysis would considerably enhance the usefulness of this kind
of simulation experiment in order to develop long-term nature conservation strategies. Moreover, a significant overlap of area potentials for agricultural activities and vascular plant
diversity was identified, indicating the high importance of large-scale conservation strategies in order to maintaining ecosystem functions. These efforts have to go hand in hand
with the development of resource-conserving agricultural management strategies with the
principal aim to produce enough food for human population.
Nevertheless, the process-based approach is a major step forward compared with the
environmental envelopes approach implemented by Ramankutty et al. (2002), since it
captures biophysical processes in more detail and allows considering a variety of crop types.
To the author’s knowledge, this simulation experiment is the first of its kind that explicitly
considers rain-fed and irrigated crop production as well as livestock grazing as separate
activities and, thus, gives an innovative and comprehensive overview of the agricultural
98
8.2. Outlook on further research
sector in Africa.
8.2. Outlook on further research
One important conclusion can be drawn from the findings summarized above: both, nature
conservation strategies and rangeland management strategies, affect the location and the
extent of agricultural area. Hence, for the development of resource-conserving agricultural
management strategies, it is important to consider both aspects.
Consequently, a possible next step would be to analyze the area potentials in the Jordan
River region under consideration of different rangeland management and nature conservation strategies. Since the Jordan River region suffers from severe water stress, the inclusion
of water availability aspects and the analysis of different agricultural management strategies on the regional freshwater resources are of high importance. Such kind of analysis
could furthermore be combined with a monetary evaluation of ecosystem services. The
implementation of such analyses, e.g. via normative scenarios, can help to maximize the
benefits from the region’s water resources for humans and ecosystems.
99
9. Bibliography
Abahussain, A. A., Abdu, A. S., Al-Zubari, W. K., El-Deen, N. A., and Abdul-Raheem,
M. (2002). Desertification in the Arab region: Analysis of current status and trends.
Journal of Arid Environments, 51(4):521–545.
Al-Jaloudy, M. (2001). FAO country pasture profile for Jordan. Online. Available at
http://www.fao.org/ag/AGP/agpc/doc/Counprof/Jordan/Jordan.htm.
Alados, C. L., ElAich, A., Papanastasis, V. P., Ozbek, H., Navarro, T., Freitas, H., Vrahnakis, M., Larrosi, D., and Cabezudo, B. (2004). Change in plant spatial patterns and
diversity along the successional gradient of Mediterranean grazing ecosystems. Ecological
Modelling, 180(4):523–535.
Alcamo, J. (2001). Scenarios as tools for international environmental assessments. Environmental issue report No. 24. European Environment Agency, Copenhagen.
Alcamo, J. (2002). Three issues for improving integrated models: Uncertainty, social
science, and legitimacy. In Gethmann, C. F. and Lingner, S., editors, Integrative Modellierung zum Globalen Wandel, pages 3–14. Springer-Verlag, Berlin, Heidelberg.
Alcamo, J., editor (2008). Environmental futures: The practice of environmental scenario
analysis. RWS Publications, Pittsburgh.
Alcamo, J., Döll, P., Henrichs, T., Kaspar, F., Lehner, B., Rösch, T., and Siebert, S.
(2003). Development and testing of the WaterGAP 2 global model of water use and
availability. Hydrological Science Journal, 48(3):317–337.
Alcamo, J., Leemans, R., and Kreileman, E., editors (1998). Global change scenarios of
the 21st century: Results from the IMAGE 2.1 model. Elsevier Science, Oxford.
Alcamo, J. and Schaldach, R. (2006). Landshift: Global modelling to assess land use
change. In Tochermann, K. and Scharl, A., editors, EnviroInfo 2006: Managing Environmental Knowledge. Proceedings of the 20th International Conference on Informatics
for Environmental Protection, pages 223–230. Shaker Verlag GmbH, Aachen.
Alexandratos, N. (1999). World food and agriculture: Outlook for the medium and longer
term. Proceedings of the National Academy of Sciences of the United States of America,
96(11):5908–5914.
101
9. Bibliography
Alhamad, M. N. (2006). Ecological and species diversity of arid Mediterranean grazing
land vegetation. Journal of Arid Environments, 66(4):698–715.
Anonymous (2009). GLOWA Jordan River scenarios of regional development under global
change. Center for Environmental Systems Research, University of Kassel, Kassel.
Antrop, M. (2000). Changing patterns in the urbanized countryside of western Europe.
Landscape Ecology, 15(3):257–270.
Arrhenius, O. (1920). Distribution of the species over the area. Meddelanden från Vetenskapsakadmiens Nobelstiftelse, 4:1–6.
Arrhenius, O. (1921). Species and area. Journal of Ecology, 9(1):95–99.
Balmford, A. and Long, A. (1994).
372(6507):623–624.
Avian endemism and forest loss.
Nature,
Barthlott, W., Hostert, A., Kier, G., Küper, W., Kreft, H., Mutke, J., Rafiqpoor, M. D.,
and Sommer, J. H. (2007). Geographic patterns of vascular plant diversity at continental
to global scales. Erdkunde, 61(4):305–315.
Bationo, A. and Buerkert, A. (2001). Soil organic carbon management for sustainable land
use in Sudano-Sahelian West Africa. Nutrient Cycling in Agroecosystems, 61(1-2):131–
142.
Bennett, E. M., Carpenter, S. R., and Caraco, N. F. (2001). Human impact on erodable
phosphorus and eutrophication: A global perspective. BioScience, 51(3):227–234.
Biggs, R., Simons, H., Bakkenes, M., Scholes, R. J., Eickhout, B., van Vuuren, D., and
Alkemade, R. (2008). Scenarios of biodiversity loss in southern Africa in the 21st century.
Global Environmental Change, 18(2):296–309.
Bondeau, A., Smith, P., Zaehle, S., Schapoff, S., Lucht, W., Cramer, W., Gerten, D.,
Lotze-Campen, H., Müller, C., and und B. Smith, M. R. (2007). Modelling the role
of agriculture for the 20th century global terrestrial carbon balance. Global Change
Biology, 13(3):679–706.
Bongaarts, J. (1996). Population pressure and the food supply system in the developing
world. Population and Development Review, 22(3):483–503.
Bousquet, F. and Le Page, C. (2004). Multi-agent simulations and ecosystem management:
A review. Ecological Modelling, 176(3–4):313–332.
Briassoulis, H. (2000). Analysis of land use change: Theoretical and modeling approaches. In Loveridge, S., editor, The Web Book of Regional Science. Online. Available at
http://www.rri.wvu.edu/WebBook/Briassoulis/Chapter5(Summary).htm.
102
Carpenter, S. R., Pingali, P. L., Bennett, E. M., and Zurek, M. B. (2005). Ecosystems and
human well-being: Scenarios. Volume 2. Island Press, Washington DC.
Ceballos, G. and Ehrlich, P. R. (2002). Mammal population losses and the extinction crisis.
Science, 296(5569):904–907.
Cecchi, G., Mattioli, R., Slingenbergh, J., and De La Rocque, S. (2008). Land cover
and tsetse fly distributions in sub-Saharan Africa. Medical and Veterinary Entomology,
22(4):364–373.
Center for International Earth Science Information Network, Columbia University,
and Centro Internacional de Agricultura Tropical (2005). Gridded population of
the world, version 3 (GPWv3): Population density grid. Online. Available at
http://sedac.ciesin.columbia.edu/gpw/.
Center for International Earth Science Information Network, Columbia University, International Food Policy Research Institute, the World Bank, and Centro Internacional de
Agricultura Tropical (2004). Global rural-urban mapping project (GRUMP): Urban/rural
population grids. Online. Available at http://sedac.ciesin.columbia.edu/gpw/.
Chapin III, F. S., Zavaleta, E. S., Eviner, V. T., Naylor, R. L., Vitousek, P. M., Reynolds,
H. L., Hooper, D. U., Lavorel, S., Sala, O. E., Hobbie, S. E., Mack, M. C., and Díaz, S.
(2000). Consequences of changing biodiversity. Nature, 405(6783):234–242.
Christensen, J., Hewitson, B., Busuioc, A., Chen, A., Gao, X., Held, I., Jones, R., Kolli, R.,
Kwon, W.-T., Laprise, R., Rueda, V., Mearns, L., Menéndez, C., Räisänen, J., Rinke, A.,
Sarr, A., and Whetton, P. (2007). Regional climate projections. In Solomon, S., Qin,
D., Manning, M., Chen, Z., Marquis, M., Averyt, K., Tignor, M., and Miller, H., editors,
Climate Change 2007: The Physical Science Basis. Contribution of Working Group I
to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, Cambridge.
CIA (2008).
The world factbook.
Online.
Available
https://www.cia.gov/library/publications/the-world-factbook/index.html.
at
Cingolani, A. M., Noy-Meir, I., and Díaz, S. (2005). Grazing effects on rangeland diversity:
A synthesis of contemporary models. Ecological Applications, 15(2):757–773.
Cromley, R. G. and Hanink, D. M. (1999). Coupling land use allocation models with raster
GIS. Journal of Geographical Systems, 1(2):137–153.
Cushman, S. (2006). Effects of habitat loss and fragmentation on amphibians: A review
and prospectus. Biological Conservation, 128(2):231–240.
Daily, G. C., editor (1997). Nature’s services: Societal dependence on natural ecosystems.
Island Press, Washington, DC.
103
9. Bibliography
de Groot, R. S., Wilson, M. A., and Boumans, R. M. (2002). A typology for the classification, description and valuation of ecosystem functions, goods and services. Ecological
Economics, 41(3):393–408.
de Leeuw, P. and Tothill, J. (1993). The concept of rangeland carrying capacity in subSaharan Africa - myth or reality. In Behnke Jr., R., Scoones, I., and Kerven, C., editors,
Range Ecology at Disequilibrium. New Models of Natural Variability and Pastoral Adaptation in African Savannas, pages 77–88. IIED, Overseas Development Institute, London.
DeFries, R. and Eshleman, K. N. (2004). Land-use change and hydrologic processes: A
major focus for the future. Hydrological Processes, 18(11):2183–2186.
Eastman, J. R., Jin, W., Kyem, P. A. K., and Toledano, J. (1995). Raster procedures
for multi-criteria/multi-objective decisions. Photogrammetric Engineering and Remote
Sensing, 61(5):539–547.
Ellis, E. C. and Ramankutty, N. (2008). Putting people in the map: Anthropogenic biomes
of the world. Frontiers in Ecology and the Environment, 6(8):439–447.
Erb, K.-H., Gaube, V., Krausmann, F., Plutzar, C., Bondeau, A., and Haberl, H. (2007).
A comprehensive global 5 min resolution land-use data set for the year 2000 consistent
with national census data. Journal of Land Use Science, 2(3):191–224.
Eswaran, H., Lal, R., and Reich, P. F. (2001). Land degradation: An overview. In Bridges,
E. M., Hannam, I. D., Oldeman, L. R., de Vries, F. W. T. P., Scherr, S., and Sompatpanit, S., editors, Responses to Land Degradation. Proceedings of the 2nd International
Conference on Land Degradation and Desertification, Khon Kaen, Thailand, pages 20–
35. Oxford Press, New Delhi, India.
Executive Action Team (1998). Overview of Middle East water resources – water resources of Palestinian, Jordanian, and Israeli interest. Online. Available at http://exactme.org/overview/index.htm.
Falkenmark, M. and Rockström, J. (2004). Balancing Water for Humans and Nature.
Earthscan, London.
FAO (2003). Review of world water resources by country. Water Reports 23. Food and
Agriculture Organization of the United Nations, Rome.
FAO (2007). Gridded livestock of the world 2007. Food and Agriculture Organization of
the United Nations, Rome.
FAO (2008). FAO statistical database. Online. Available at http://faostat.fao.org/.
FAO (2009). FAO statistical database. Online. Available at http://faostat.fao.org/.
104
Field, C. B., Campbell, J. E., and Lobell, D. B. (2008). Biomass energy: The scale of the
potential resource. Trends in Ecology and Evolution, 23(2):65–72.
Fischer, G., Teixeira, E., Hizsnik, E., and van Velthuizen, H. (2008). Land use dynamics and
sugarcane production. In Zuurbier, P. and van de Vooren, J., editors, Sugarcane ethanol, Contributions to climate mitigation and the environment. Wageningen Academic
Publishers, Wageningen.
Fischer, G., van Velthuizen, H., Shah, M., and Nachtergaele, F. (2002). Global agroecological assessment for agriculture in the 21st century: Methodology and results. RR02-02. International Institute for Applied Systems Analysis, Laxenburg.
Foley, J. A., DeFries, R., Asner, G. P., Barford, C., Bonan, G., Carpenter, S. R., Chapin,
F. S., Coe, M. T., Daily, G. C., Gibbs, H. K., Helkowski, J. H., Holloway, T., Howard,
E. A., Kucharik, C. J., Monfreda, C., Patz, J. A., Prentice, I. C., Ramankutty, N., and
Snyder, P. K. (2005). Global consequences of land use. Science, 309(5734):570–574.
Geist, H. and Lambin, E. (2004). Dynamic causal patterns of desertification. BioScience,
54(9):817–829.
Geist, H. J. and Lambin, E. F. (2001). What drives tropical deforestation? A meta-analysis
of proximate and underlying causes of deforestation based on subnational case study
evidence. LUCC Report Series No. 4. LUCC International Project Office.
Geist, H. J. and Lambin, E. F. (2002). Proximate causes and underlying driving forces of
tropical deforestation. BioScience, 52(2):134–150.
Geneletti, D. (2004). A GIS-based decision support system to identify nature conservation
priorities in an alpine valley. Land Use Policy, 21(2):149–160.
Gillson, L. and Hoffman, M. (2007). Rangeland ecology in a changing world. Science,
315(5808):53–54.
Grimm, V., Berger, U., Bastiansen, F., Eliassen, S., Ginot, V., Giske, J., Goss-Custard,
J., Grand, T., Heinz, S. K., Huse, G., Huth, A., Jepsen, J. U., Jørgensen, C., Mooij,
W. M., Müller, B., Pe’er, G., Piou, C., Railsback, S. F., Robbins, A. M., Robbins, M. M.,
Rossmanith, E., Rüger, N., Strand, E., Souissi, S., Stillman, R. A., Vabø, R., Visser, U.,
and DeAngelis, D. L. (2006). A standard protocol for describing individual-based and
agent-based models. Ecological Modelling, 198(1-2):115–126.
Haberl, H., Erb, K., and Krausmann, F. (2007a). Human appropriation of net primary
production (HANPP). Internet Encyclopaedia of Ecological Economics. Online. Available
at http://www.ecoeco.org/pdf/2007_march_hanpp.pdf.
105
9. Bibliography
Haberl, H., Erb, K., Krausmann, F., Gaube, V., Bondeau, A., Plutzar, C., Gingrich, S.,
Lucht, W., and Fischer-Kowalski, M. (2007b). Quantifying and mapping the human
appropriation of net primary production in Earth’s terrestrial ecosystems. Proceedings
of the National Academy of Sciences of the United States of America, 104(31):12942–
12947.
Hall, C. A. S., Tian, H., Qi, Y., and Cornell, G. P. J. (1995). Modelling spatial and temporal
patterns of tropical land-use change. Journal of Biogeography, 22(4–5):753–757.
Harrison, P., Berry, P., Butt, N., and New, M. (2006). Modelling climate change impacts
on species’ distributions at the European scale: Implications for conservation policy.
Environmental Science and Policy, 9(2):116–128.
Heistermann, M. (2006). Modelling the Global Dynamics of Rain-fed and Irrigated Croplands. PhD thesis, University of Kassel, Germany.
Heistermann, M., Müller, C., and Ronneberger, K. (2006). Land in sight? Achievements,
deficits and potentials of continental to global scale land-use modeling. Agriculture,
Ecosystems and Environment, 114(2–4):141–158.
Hilferink, M. and Rietveld, P. (1999). LAND USE SCANNER: An integrated GIS based model for long term projections of land use in urban and rural areas. Journal of
Geographical Systems, 1(2):155–177.
Hoogwijk, M., Faaij, A., Eickhout, B., de Vries, B., and Turkenburg, W. (2005). Potential
of biomass energy out to 2100, for four IPCC SRES land-use scenarios. Biomass and
Bioenergy, 29(4):225–257.
Houghton, J. T., Ding, Y., Griggs, D. J., Noguer, M., van der Linden, P. J., Dai, X.,
Maskell, K., and Johnson, C. A., editors (2001). Climate Change 2001: The Scientific
Basis. Cambridge University Press, Cambridge.
Hughes, B. and Hillebrand, E. (2006). Exploring and Shaping International Futures. Paradigm Publishers, Boulder, Colorado USA.
Hughes, B. B. (1999). The International Futures (IFs) modeling project. Simulation and
Gaming, 30(3):304–326.
Ibáñez, J., Martínez, J., and Schnabel, S. (2007). Desertification due to overgrazing in
a dynamic commercial livestock-grass-soil system. Ecological Modelling, 205(3-4):277–
288.
IPCC (2007). Climate Change 2007: Synthesis Report. Contribution of Working Groups I,
II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Intergovernmental Panel on Climate Change, Geneva, Switzerland.
106
IPCC (2007). Climate Change 2007: The Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report of the IPCC. Cambridge University Press,
Cambridge, 996 pp.
Kaimowitz, D. and Angelsen, A., editors (1998). Economic Models of Tropical Deforestation: A Review. Center for International Forestry Research, Bogor, Indonesia.
Kalnay, E. and Cai, M. (2003). Impact of urbanization and land-use change on climate.
Nature, 423(6939):528–531.
Kan, I., Rapaport-Rom, M., and Shechter, M. (2007). Economic analysis of climate-change
impacts on agricultural profitability and land use: The case of Israel. In 15th Annual
Conference of the European Association of Environmental and Resource Economists
(EAERE), Thessaloniki, Greece.
Klein Goldewijk, K. (2001). Estimating global land use change over the past 300 years:
the HYDE database. Global Biogeochemical Cycles, 15(2):417–433.
Klein Goldewijk, K. (2005). Three centuries of global population growth: A spatial referenced population (density) database for 1700 - 2000. Population and Environment,
26(4):343–367.
Koch, J., Schaldach, R., and Köchy, M. (2008). Modeling the impacts of grazing land
management on land-use change for the jordan river region. Global and Planetary
Change, 64(3-4):177–187.
Koch, J., Schaldach, R., Onigkeit, J., Ceglarek, T., Alcamo, J., Köchy, M., Wolff, H.-P.,
and Kan, I. Future land-use and land-cover change scenarios for the Jordan River region.
Ecology and Society. In review.
Köchy, M. (2006). Stochastic time series of daily precipitation for the interior of Israel.
Israel Journal of Earth Sciences, 55(2):103–109.
Köchy, M. (2007). Grazing capacity of Middle East landscapes under contrasting climate
change scenarios. In 37. Jahrestagung der Gesellschaft für Ökologie, volume 37:137,
Marburg, Germany. Verhandlungen der Gesellschaft für Ökologie.
Köchy, M., Mathaj, M., Jeltsch, F., and Malkinson, D. (2008). Resilience of stocking capacity to changing climate in arid to Mediterranean landscapes. Regional Environmental
Change, 8(2):73–87.
Küper, W., Sommer, J. H., Lovett, J. C., Mutke, J., Linder, H. P., Beentje, H. J., Van
Rompaey, R. S. A. R., Chatelain, C., Sosef, M., and Barthlott, W. (2004). Africa’s
hotspots of biodiversity redefined. Annals of the Missouri Botanical Garden, 91(4):525–
535.
107
9. Bibliography
Kremen, C., Williams, N. M., and Thorp, R. W. (2002). Crop pollination from native
bees at risk from agricultural intensification. Proceedings of the National Academy of
Sciences of the United States of America, 99(26):16812–16816.
Kruess, A. and Tscharntke, T. (1994). Habitat fragmentation, species loss, and biological
control. Science, 264(5165):1581–1584.
Kuhnert, M., Voinov, A., and Seppelt, R. (2005). Comparing raster map comparison
algorithms for spatial modeling and analysis. Photogrammetric Engineering and Remote
Sensing, 71(8):975–984.
Lambin, E. F. (1997). Modelling and monitoring land-cover change processes in tropical
regions. Progress in Physical Geography, 21(3):375–393.
Lambin, E. F. and Geist, H., editors (2006). Introduction: Local processes with global
impacts. Springer-Verlag, Berlin, Heidelberg.
Lambin, E. F. and Geist, H. J. (2007).
Causes of land-use and land-cover
change. In Cleveland, C. J., editor, Encyclopedia of Earth. Online. Available at
http://www.eoearth.org/article/Causes_of_land-use_and_land-cover_change.
Lambin, E. F., Geist, H. J., and Lepers, E. (2003). Dynamics of land-use and land-cover
change in tropical regions. Annual Review of Environment and Resources, 28:205–241.
Lambin, E. F., Rounsevell, M. D. A., and Geist, H. J. (2000). Are agricultural landuse models able to predict changes in land-use intensity? Agriculture, Ecosystems and
Environment, 82(1-3):321–331.
Loveland, T., Reed, B., Brown, J., Ohlen, D., Zhu, Z., Yang, L., and Merchant, J. (2000).
Development of a global land cover characteristics database and IGBP DISCover from
1 km AVHRR data. International Journal of Remote Sensing, 21(6–7):1303–1330.
Luck, G. W., Ricketts, T. H., Daily, G. C., and Imhoff, M. (2004). Alleviating spatial conflict
between people and biodiversity. Proceedings of the National Academy of Sciences of
the United States of America, 101(1):182–186.
Malkinson, D. and Jeltsch, F. (2007). Intraspecific facilitation: a missing process along
increasing stress gradients insights from simulated shrub populations. Ecography,
30(3):339–348.
Martino, D. and Zommers, Z. (2007). Chapter 1: Environment for Development. In United
Nations Environment Programme, editor, Global Environmental Outlook GEO4, pages
1–36. Earth Print Limited, Stevenage, Hertfordshire.
Matson, P. A., Parton, W. J., Power, A. G., and Swift, M. J. (1997). Agricultural intensification and ecosystem properties. Science, 277(5325):504–509.
108
McGarigal, K., Cushman, S., Neel, M., and Ene, E. (2002). FRAGSTATS: Spatial
pattern analysis program for categorical maps. Computer software program produced
by the authors at the University of Massachusetts, Amherst. Online. Available at
http://www.umass.edu/landeco/research/fragstats/fragstats.html.
Menzel, L., Koch, J., Onigkeit, J., and Schaldach, R. (2009). Modelling the effects of landuse and land-cover change on water availability in the Jordan River region. Advances in
Geosciences, 21:73–80.
Metzger, M. J., Rounsevell, M. D. A., Acosta-Michlik, L., Leemans, R., and Schröter,
D. (2006). The vulnerability of ecosystem services to land use change. Agriculture,
Ecosystems and Environment, 114(1):69–85.
Meyer, W. B. and Turner II, B. L. (1994). Changes in land use and land cover: A global
perspective. Cambridge University Press, Cambridge.
Millennium Ecosystem Assessment (2003). Ecosystems and Human Well-Being: A Framework For Assessment. Island Press, Washington, DC.
Millennium Ecosystem Assessment (2005a). Ecosystems and Human Well-being: Desertification Synthesis, Millennium Ecosystem Assessment Series. Island Press, Washington,
DC.
Millennium Ecosystem Assessment (2005b). Ecosystems and Human Well-being: Scenarios,
Volume 2: Findings of the Scenario Working Group, Millennium Ecosystem Assessment
Series. Island Press, Washington, DC.
Miller, E. J., Kriger, D. S., and Hunt, J. D. (1999). TCRP Web Document 9: Integrated
Urban Models for Simulation of Transit and Land-Use Policies - Final Report. University
of Toronto Joint Program in Transportation and DELCAN Corporation, Toronto.
Ministry of Environment of The Hashemite Kingdom of Jordan (2007).
Third country report on the implementation of the United Nations convention to combat desertification (UNCCD).
Online.
Available at
http://www.unccd.int/cop/reports/asia/national/2007/jordaneng.pdf.
Mitchell, T. D. and Jones, P. D. (2005). An improved method of constructing a database of
monthly climate observations and associated high-resolution grids. International Journal
of Climatology, 25(6):693–712.
Msangi, S. and Rosegrant, M. (2009). World agriculture in a dynamically-changing environment: IFPRI’s long-term outlook for food and agriculture under additional demand and constraints. Online. Available at http://www.fao.org/wsfs/forum2050/wsfsbackground-documents/wsfs-expert-papers/en/.
109
9. Bibliography
Mutke, J. and Barthlott, W. (2005). Patterns of vascular plant diversity at continental to
global scales. Biologiske Skrifter, 55:521–537.
Myers, N. and Kent, J. (2003). New consumers: The influence of affluence on the environment. Proceedings of the National Academy of Sciences of the United States of
America, 100(8):4963–4968.
NIMA (1997).
Vector map level 0 (VMap0).
Mapping Agency (NIMA), Washington, DC.
www.mapability.com/info/vmap0_index.html.
National Imagery and
Online. Available at
Nordblom, T., Goodchild, A., Shomo, F., and Gintzburger, G. (1997). Dynamics of feed
resources in mixed farming systems of west/central asia-north africa. In Crop Residues
in Sustainable Mixed Crops/Livestock Farming Systems. International Crops Research
Institute for the Semi-Arid Tropics (ICRISAT), India.
Norris, K. (2008). Agriculture and biodiversity conservation: Opportunity knocks. Conservation Letters, 1(1):2–11.
Oehl, F., Sieverding, E., Ineichen, K., Mäder, P., Boller, T., and Wiemken, A. (2003). Impact of land use intensity on the species diversity of arbuscular mycorrhizal fungi in agroecosystems of central Europe. Applied and Environmental Microbiology, 69(5):2816–
2824.
Onigkeit, J., Lübkert, B., and Alcamo, J. (2007). Quantitative model drivers for the
GLOWA Jordan River scenarios. Technical report. Center for Environmental Systems
Research, University of Kassel, Kassel.
Orme, C. D. L., Davies, R. G., Burgess, M., Eigenbrod, F., Pickup, N., Olson, V. A., Webster, A. J., Ding, T.-S., Rasmussen, P. C., Ridgely, R. S., Stattersfield, A. J., Bennett,
P. M., Blackburn, T. M., Gaston, K. J., and Owens, I. P. F. (2005). Global hotspots of
species richness are not congruent with endemism or threat. Nature, 436(7053):1016–
1019.
Parker, D. C., Manson, S. M., Janssen, M. A., Hoffmann, M. J., and Deadman, P. (2003).
Multi-agent systems for the simulation of land-use and land-cover change: A review.
Annals of the Association of American Geographers, 93(2):314–337.
Parry, M. L., Rosenzweig, C., Iglesias, A., Livermore, M., and Fischer, G. (2003). Effects
of climate change on global food production under SRES emissions and socio-economic
scenarios. Global Environmental Change, 14(1):53–67.
Parton, W. J., Hartman, M., Ojima, D., and Schimel, D. (1998). DayCent and its land
surface submodel: Description and testing. Global and Planetary Change, 19(1):35–48.
110
Peel, M., Finlayson, B., and McMahon, T. (2007). Updated world map of the KöppenGeiger climate classification. Hydrology and Earth System Sciences, 11(5):1633–1644.
Perevolotsky, A. and Landau, S. (1992). Droits pastoraux en israel: Perspectives historique
et ecologique sur le statut des terres de parcours. In Terres collectives en Mediterranee.
Histoire, legislation, usages et modes d’utilisation par les animaux, pages 117–135. Food
and Agriculture Organization of the United Nations, Rome.
Perevolotsky, A., Landau, S., Kababia, D., and Ungar, E. (1998). Diet selection in dairy goats grazing woody Mediterranean rangeland. Applied Animal Behaviour Science,
57(1):117–131.
Pielke Sr, R. A., Marland, G., Betts, R. A., Chase, T. N., Eastman, J. L., Niles, J. O., Niyogi,
D. E. S., and Running, S. W. (2002). The influence of land-use change and landscape
dynamics on the climate system: Relevance to climate-change policy beyond the radiative
effect of greenhouse gases. Royal Society of London Philosophical Transactions A,
360(1797):1705–1719.
Pimm, S. L. and Raven, P. (2000). Extinction by numbers. Nature, 403(6772):843–845.
Pontius Jr., R. G. (2002). Statistical methods to partition effects of quantity and location during comparison of categorical maps at multiple resolutions. Photogrammetric
Engineering and Remote Sensing, 68(10):1041–1049.
Pontius Jr., R. G., Huffaker, D., and Denman, K. (2004). Useful techniques of validation
for spatially explicit land-change models. Ecological Modelling, 179(4):45–461.
Pontius Jr., R. G. and Schneider, L. (2001). Land-cover change model validation by an
ROC method for the Ipswich watershed, Massachusetts, USA. Agriculture, Ecosystems
and Environment, 85(1–3):239–248.
Ramankutty, N., Evan, A. T., Monfreda, C., and Foley, J. A. (2008). Farming the planet: 1. Geographic distribution of global agricultural lands in the year 2000. Global
Biogeochemical Cycles, 22:GB1003.
Ramankutty, N., Foley, J., Norman, J., and McSweeney, K. (2002). The global distribution
of cultivable lands: Current patterns and sensitivity to possible climate change. Global
Ecology and Biogeography, 11(5):377–392.
Rosegrant, M., Cai, X., and Cline, S. (2002a). World water and food to 2025: Dealing
with scarcity. International Food Policy Research Institute, Washington, DC. 348 pp.
Rosegrant, M., Cline, S. A., Li, W., Sulser, T. B., and Valmonte-Santos, R. (2005a). Looking ahead: Long-term prospects for Africa’s agricultural development and food security.
Discussion Paper No. 41. International Food Policy Research Institute, Washington, DC.
111
9. Bibliography
Rosegrant, M., Meijer, S., and Cline, S. (2002b). International model for policy analysis of
agricultural commodities and trade (IMPACT): Model description. International Food
and Policy Research Institute, Washington, DC.
Rosegrant, M. W., Ringler, C., Msangi, S., Cline, S. A., and Sulser, T. B. (2005b). International model for policy analysis of agricultural commodities and trade (IMPACTWATER): Model description. International Food Policy Research Institute, Washington,
DC.
Rosegrant, M. W., Ringler, C., Msangi, S., Sulser, T. B., Zhu, T., and Cline, S. A. (2008).
International model for policy analysis of agricultural commodities and trade (IMPACT):
Model description. International Food and Policy Research Institute, Washington, DC.
Rothman, D. S., Agard, J., and Alcamo, J. (2007). Chapter 9: The future today. In United
Nations Environment Programme), a. . S., editor, Global Environmental Outlook GEO4,
pages 395–454. Earth Print Limited.
Rouget, M., Richardson, D. M., Cowling, R. M., Lloyd, J. W., and Lombard, A. T. (2003).
Current patterns of habitat transformation and future threats to biodiversity in terrestrial
ecosystems of the Cape Floristic Region, South Africa. Biological Conservation, 112(1–
2):63–85.
Rykiel Jr., E. J. (1996). Testing ecological models: The meaning of validation. Ecological
Modelling, 90(3):229–244.
Saaty, T. L. (1996). Multicriteria decision making. Elsevier, Amsterdam.
Sala, O., Chapin III, 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., LeRoy Poff, N., Sykes, M., Walker, B., Walker, M., and Wall, D. (2000).
Global biodiversity scenarios for the year 2100. Science, 287(5459):1770–1774.
Scanlon, B. R., Jolly, I., Sophocleous, M., and Zhang, L. (2007). Global impacts of
conversions from natural to agricultural ecosystems on water resources: Quantity versus
quality. Water Resources Research, 43:W03437.
Schaldach, R. and Alcamo, J. (2006). Coupled simulation of regional land use change and
soil carbon sequestration: A case study for the state of Hesse in Germany. Environmental
Modelling and Software, 21(10):1430–1446.
Schaldach, R., Alcamo, J., and Heistermann, M. (2006). The multiple scale land use
change model LandSHIFT: A scenario analysis of land use change and environmental
consequences in Africa. In Voinov, A., Jakeman, A., and Rizzoli, A., editors, Proceedings
of the iEMSs Third Biennal Meeting: “Summit on Environmental Modelling and Software”. International Environmental Modelling and Software Society, Burlington, USA.
112
Schaldach, R. and Koch, J. (2009). Conceptual design and implementation of a model for
the integrated simulation of large-scale land-use systems. In Athanasiadis, I., Mitkas,
P., Rizzoli, A., and Marx-Gómez, J., editors, Information Technologies in Environmental
Engineering, pages 425–438. Springer Verlag, Berlin, Heidelberg.
Schaldach, R. and Priess, J. A. (2008). Integrated models of the land system: A review
of modelling approaches on the regional to global scale. Living Reviews in Landscape
Research, 2. Available at http://landscaperesearch.livingreviews.org/Articles/lrlr-20081/.
Schaldach, R., Priess, J. A., and Alcamo, J. Simulating the impact of bio-fuel development
on country-wide land-use change in India. Biomass and Bioenergy. In review.
Schneider, L. C. and Pontius Jr., R. G. (2001). Modeling land-use change in the Ipswich
watershed, Massachusetts, USA. Agriculture, Ecosystems and Environment, 85(1–3):83–
94.
Scholes, R. J. and Biggs, R. (2005). A biodiversity intactness index. Nature, 434(7029):45–
49.
Schreinemachers, P. and Berger, T. (2006). Land use decisions in developing countries and
their representation in multi-agent systems. Journal of Land Use Science, 1(1):29–44.
Seré, C. and Steinfeld, H. (1996). World livestock production systems - current status,
issues and trends. FAO Animal Production and Health Paper 127. Food and Agriculture
Organization of the United Nations, Rome.
Sitch, S., Smith, B., Prentice, C., Arneth, A., Bondeau, A., Cramer, W., Kaplans, J.,
Levis, S., Lucht, W., Sykes, M., Thonicke, K., and Venevsky, S. (2003). Evaluation of
ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic
global vegetation model. Global Change Biology, 9(2):161–185.
Solecki, W. D. and Oliveri, C. (2004). Downscaling climate change scenarios in an urban
land use change model. Journal of Environmental Management, 72(1-2):105–115.
Stehfest, E., Heistermann, M., Priess, J. A., Ojima, D. S., and Alcamo, J. (2007). Simulation of global crop production with the ecosystem model DayCent. Ecological Modelling,
209(2-4):203–219.
Steinfeld, H., de Haan, C., and Blackburn, H. (1998). Livestock and the environment:
Issues and options. In Lutz, E., editor, Agriculture and the environment: Perspectives
on sustainable rural development, pages 283–302. The World Bank, Washington, DC.
Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., Rosales, M., and de Haan, C. (2006).
Livestock’s long shadow. Food and Agriculture Organization of the United Nations,
Rome.
113
9. Bibliography
Stéphenne, N. and Lambin, E. (2001). A dynamic simulation model of land-use changes in
Sudano-sahelian countries of Africa (SALU). Agriculture, Ecosystems and Environment,
85(1-3):145–161.
Tilman, D. (1998). The greening of the green revolution. Nature, 396(6708):211–212.
Tilman, D. (1999). Global environmental impacts of agricultural expansion: The need for
sustainable and efficient practices. Proceedings of the National Academy of Sciences of
the United States of America, 96(11):5995–6000.
Tilman, D., Cassman, K. G., Matson, P. A., Naylor, R., and Polasky, S. (2002). Agricultural
sustainability and intensive production practices. Nature, 418(6898):671–677.
Tscharntke, T., Klein, A. M., Kruess, A., Steffan-Dewenter, I., and Thies, C. (2005).
Landscape perspectives on agricultural intensification and biodiversity - ecosystem service management. Ecology Letters, 8(8):857–874.
Tsioras, P. A., Tsiourlis, G., Papanastasis, V. P., Casimiro, P. C., and Xanthopoulos, G.
(2006). A report on trends and relationships between climate changes, livestock husbandry, agricultural activities, fire and desertification in marginal lands. Deliverable of
the DeSurvey Research Project, Module “Rangeland, Marginal Land and Forest Vulnerability”. Available at http://www.nateko.lu.se/desurvey10/.
Tubiello, F. N., Soussana, J.-F., and Howden, S. M. (2007). Crop and pasture response to
climate change. Proceedings of the National Academy of Sciences of the United States
of America, 104(50):19686–19690.
Turner II, B. L., Skole, D., Sanderson, S., Fischer, G., Fresco, L., and Leemans, R. (1995).
Land-use and land-cover change science/research plan. IGBP Global Change Report
35/HDP Report 7. International Geosphere-Biosphere Programme, Human Dimensions
Global Environmental Change Programme, Stockholm/Geneva.
UNCCD (1994). United Nations convention to combat desertification in countries experiencing serious drought and/or desertification, particularly in Africa. United Nations
Environmental Programme, Nairobi.
United Nations (2009a). World population prospect: The 2008 revision. Online. Available
at http://esa.un.org/unpp/.
United Nations (2009b). World urbanization prospect: The 2005 revision. Online. Available
at http://www.un.org/esa/population/publications/WUP2005/2005wup.htm.
U.S. EPA (2000). Projecting land-use change: A summary of models for assessing the
effects of community growth and change on land-use patterns. United States Environmental Protection Agency, Office of Research and Development, Cincinnati.
114
USGS (1998).
HYDRO1k: Elevation derivative database.
United States
Geological Survey EROS Data Center, Sioux Falls.
Online. Available at
http://edc.usgs.gov/products/elevation/gtopo30/hydro/index.html.
Uuemaa, E., Antrop, M., Roosaare, J., Marja, R., and Mander, . (2009). Landscape metrics
and indices: An overview of their use in landscape research. Living Reviews in Landscape
Research, 3. Available at http://www.livingreviews.org/lrlr-2009-1.
van de Koppel, J. and Rietkerk, M. (2000). Herbivore regulation and irreversible vegetation
change in semi-arid grazing systems. Oikos, 90(2):253–260.
van de Koppel, J., Rietkerk, M., van Langevelde, F., Kumar, L., Klausmeier, C. A., Fryxell,
J. M., Hearne, J. W., van Andel, J., de Ridder, N., Skidmore, A., Stroosnijder, L.,
and Prins, H. H. T. (2002). Spatial heterogeneity and irreversible vegetation change in
semiarid grazing systems. The American Naturalist, 159(2):209–218.
Veldkamp, A. and Fresco, L. O. (1996). CLUE: A conceptual model to study the conversion
of land use and its effects. Ecological Modelling, 85(2–3):253–270.
Verburg, P., Kok, K., Pontius Jr., R., and Veldkamp, A. (2006). Modeling land-use and
land-cover change. In Lambin, E. F. and Geist, H. J., editors, Land-use and land-cover
change. Local processes and global impacts, pages 117–135. Springer, Berlim.
Verburg, P. H., Schot, P. P., Dijst, M. J., and Veldkamp, A. (2004). Land use change
modelling: Current practice and research priorities. GeoJournal, 61(4):309–324.
Verburg, P. H., Soepboer, W., Veldkamp, A., Limpiada, R., Espaldon, V., and Mastura, S.
S. A. (2002). Modeling the spatial dynamics of regional land use: The CLUE-S model.
Environmental Management, 30(3):391–405.
von Thünen, J. H. (1966). Der isolierte Staat in Beziehung auf Landwirtschaft und Nationalökonomie. Wissenschaftliche Buchgesellschaft, Darmstadt, Germany. Nachdruck der
3. Auflage von 1875.
Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesank, C., Beebee, T. J. C., Fromentin, J.-M., Hoegh-Guldberg, O., and Bairlein, F. (2002). Ecological responses to
recent climate change. Nature, 416(6879):389–395.
WBGU (2008). Welt im Wandel - Zukunftsfähige Bioenergie und nachhaltige Landnutzung.
Wissenschaftlicher Beirat der Bundesregierung Globale Umweltveränderungen, Berlin. pp
388.
WDPA Consortium (2004). World database on protected areas. Online. Available at
http://www.unep-wcmc.org/wdpa/.
115
9. Bibliography
Weiß, M., Schaldach, R., Alcamo, J., and Flörke, M. (2009). Quantifying the human
appropriation of fresh water by African agriculture. Ecology and Society, 14(2):25.
White, R. and Engelen, G. (1997). Cellular automata as the basis of integrated dynamic
regional modelling. Environment and Planning B: Planning and Design, 24(2):235–246.
Wint, W., Chilonda, P., Gerber, P., Robinson, T., and Wassenaar, T. (2003). Defining
land unsuitable for livestock. Food and Agriculture Organization of the United Nations,
Rome.
Wint, W. and Rogers, D. (2000). Predicted distributions of Tsetse in Africa. Consultancy report for the Animal Health Service of the Animal Production and
Health Division of the Food and Agriculture Organization of the United Nations.
Food and Agriculture Organization of the United Nations, Rome. Online. Available at
http://www.fao.org/ag/againfo/programmes/en/paat/documents/maps/pdf/tserep.pdf.
Wolfram, S. (1984). Universality and complexity in cellular automata. Physica D: Nonlinear
Phenomena, 10(1–2):1–35.
Wood, S., Sebastian, K., and Scherr, S. J. (2000). Pilot analysis of global ecosystems:
Agroecosystems. International Food Policy Research Institute and World Resources
Institute, Washington, DC.
Wu, F. (1999). GIS-based simulation as an exploratory analysis for space-time processes.
Journal of Geographical Systems, 1(3):199–218.
Zechmeister, H. and Moser, D. (2001). The influence of agricultural land-use intensity on
bryophyte species richness. Biodiversity and Conservation, 10(10):1609–1625.
116
A. Nonlinear correlation functions
between stocking density and
biomass productivity
117
Fig. A.1.: Non-linear correlation functions between biomass productivity and stocking density, calculated with WADISCAPE.
A. Nonlinear correlation functions between stocking density and biomass productivity
118
B. Input specification
119
120
1992/1993
2000
2000, 2050
2000, 2050
Land use/cover
Population density
Biomass productivity
Stocking capacity
Slope
Infrastructure
River network density
Conservation area
Micro
Intermediate
Temporal coverage
1999-2001
1999-2001
1999-2001
2000
2000
2025,2050
2025,2050
2025,2050
2025,2050
1961-1990
Model variable
Crop production
Human population
Livestock number
Area harvested
Irrigated area
Human population
Livestock number
Crop production
Yield increase
Wheat yield
Spatial level
Country
Biomass productivity
Biomass productivity
Preference ranking
Preference ranking
Preference ranking
Land-use constraint
Baseline definition
Initial condition
Purpose
Baseline definition
Baseline definition
Baseline definition
Baseline definition
Baseline definition
Scenario driver
Scenario driver
Scenario driver
Scenario driver
Biomass productivity
WADISCAPE results
WADISCAPE results
Based on HYDRO1k data set
Based on VMap0 data set on roads
Based on HYDRO1k data set on streams
(Inter-)National conservation areas
Based on IMPACT results
Based on IMPACT results
Based on IMPACT results
Yield distribution influenced by climate,
soil, and management
Land cover classification
GRUMP alpha
Prepared for simulation experiment
Comment
Stehfest et al. (2007)
Loveland et al. (2000)
Center for International
Earth Science Information
Network et al. (2004)
Köchy et al. (2008)
Köchy et al. (2008)
USGS (1998)
NIMA (1997)
USGS (1998)
WDPA Consortium (2004)
MEA (2005b)
Ringler (pers. comm.)
Ringler (pers. comm.)
Ringler (pers. comm.)
Source
FAO (2008)
FAO (2008)
FAO (2008)
FAO (2008)
Tab. B.1.: Summary of data requirements for the simulation experiment on feedback effects between stocking density and
green biomass production (chapter 4).
B. Input specification
121
Stocking capacity
Biomass productivity
Slope
Infrastructure
River network density
Conservation area
Crop production (Israel)
Land use/cover
Population density
Ecoregion
Micro
Intermediate
Model Variable
Crop production
Human population
Livestock number
Area harvested
Irrigated area
Permanent meadows
Human population
Livestock number
Crop production (Jordan/PA)
Yield change (Jordan/PA)
Wheat yield
Spatial level
Country
1961-1990
1961-1990
2010,2030,2050
1992/1993
1990
Temporal coverage
1999-2001
1999-2001
1999-2001
1999-2001
1999-2001
1999-2001
2010,2030,2050
2010,2030,2050
2010,2030,2050
2010,2030,2050
1961-1990
Biomass productivity
Biomass productivity
Preference ranking
Preference ranking
Preference ranking
Land-use constraint
Scenario driver
Baseline definition
Preference ranking
Purpose
Baseline definition
Baseline definition
Baseline definition
Baseline definition
Baseline definition
Baseline definition
Scenario driver
Scenario driver
Scenario driver
Scenario driver
Biomass productivity
WADISCAPE results
WADISCAPE results
HYDRO1k
Road/railroad density, VMap0
VMap0
(Inter-)National conservation areas
Based on IMPACT results
Based on IMPACT results
Based on IMPACT results
Yield distribution influenced by climate,
soil, and management
VALUE results
Land cover classification
GRUMP alpha
Comment
Stehfest et al. (2007)
Kan et al. (2007)
Loveland et al. (2000)
Center for International
Earth Science Information
Network et al. (2004)
Köchy et al. (2008)
Köchy et al. (2008)
USGS (1998)
NIMA (1997)
NIMA (1997)
WDPA Consortium (2004)
Source
FAO (2009)
FAO (2009)
FAO (2009)
FAO (2009)
FAO (2009)
FAO (2009)
SAS
SAS
SAS
SAS
Tab. B.2.: Summary of data requirements for the simulation experiment on land-use and land-cover change scenarios for the
Jordan River region (chapter 6).
C. Land-use and land-cover maps
for the simulation experiment
on future conflicts between
agricultural land use and
biodiversity in Africa
123
Fig. C.1.: Land-use and land-cover map for 2020 (left) and map of changes in land use and land cover as compared to baseline
(right) - simulation run one.
C. Land-use and land-cover maps
124
125
Fig. C.2.: Land-use and land-cover map for 2020 (left) and map of changes in land use and land cover as compared to baseline
(right) - simulation run two.
Fig. C.3.: Land-use and land-cover map for 2020 (left) and map of changes in land use and land cover as compared to baseline
(right) - simulation run three.
C. Land-use and land-cover maps
126
127
Fig. C.4.: Land-use and land-cover map for 2020 (left) and map of changes in land use and land cover as compared to baseline
(right) - simulation run four.
ISBN 978-3-89958-964-1
Modeling the impacts of land-use change on ecosystems at the regional and continental scale
Jennifer Koch
Modeling the impacts of land-use
change on ecosystems at the
regional and continental scale
Jennifer Koch