Subscriber access provided by Northern Illinois University
Policy Analysis
A global synthesis of Jatropha cultivation: Insights into land use change and management practices. David Christopher Walmsley, Rob Bailis, and Alexandra-Maria Klein Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.6b01274 • Publication Date (Web): 10 Aug 2016 Downloaded from http://pubs.acs.org on August 16, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 37
Environmental Science & Technology
1
A global synthesis of Jatropha cultivation: Insights into land use change and management
2
practices.
3
4
David C. Walmsley1*, Rob Bailis2, Alexandra-Maria Klein3
5
6
1
7
1, 21335, Lüneburg, Germany
8
2
Stockholm Environment Institute, 11 Curtis Ave, Somerville MA 02144, USA
9
3
Nature Conservation and Landscape Ecology, University of Freiburg, Tennenbacher Str. 4,
10
Institute of Ecology, Faculty of Sustainability, Leuphana University Lüneburg, Scharnhorst Str.
79106 Freiburg, Germany
11
12 13
*Corresponding author: D. Walmsley, Email:
[email protected], phone: +49 4131 677
14
2083, fax: +49 4131 677 2849
15 16
1 ACS Paragon Plus Environment
Environmental Science & Technology
Page 2 of 37
17
Abstract
18
Despite setbacks interest in Jatropha cultivation remains high. This study addressed the question
19
to what extent Jatropha cultivation has replaced specific vegetation and land use types and how
20
the existing areas are managed. Major forms of land use change and management practices were
21
identified based on cluster analysis of data from 106 interviewee’s responses to a comprehensive
22
global survey. Of the 1.04 ×106 ha cultivated with Jatropha in 2011 40% were established on land
23
that was cleared of vegetation as a result of logging activities unrelated to Jatropha cultivation,
24
34% was defined as unused, and the remainder was attributable to areas previously used for crops
25
or animal husbandry. With the exception of croplands, these areas were dominated (90-98%) by a
26
few internationally active companies whose cultivation models were almost exclusively based on
27
outgrower schemes. Management practices were largely extensive in nature (low mechanical
28
input and infrequent use of fertilizers, pesticides and herbicides), and also dominated by large
29
projects. Broad surveys such as this are useful in identifying general trends in this emerging
30
global industry, but detailed case studies, particularly of large projects, are needed in order to
31
draw more informed conclusions about the site-specific impacts of Jatropha cultivation.
32
33 34
Introduction
35
Diminishing natural resources and the frequently negative impacts of current resource acquisition
36
create the need for more sustainable resource avenues. The use of biomass as a non-food
37
commodity, particularly as a substitute for fossil fuel, is therefore being increasingly promoted.1–3
38
Many concepts utilizing different sources of organic matter have been suggested4 and are
39
receiving increased interest on a commercial scale.5,6 However, increased biofuel production 2 ACS Paragon Plus Environment
Page 3 of 37
Environmental Science & Technology
40
raised concerns about ecological impacts and the extent to which biofuel crops either displaced
41
food production or land providing important ecosystem functions.
42
With strong associations between common oilseed feedstocks and deforestation, such as soy in
43
the Brazilian Amazon7 and oil palm in Indonesia,8 and emerging policies to encourage or
44
mandate more sustainable alternatives,9 the biofuel industry sought novel crops. Of the many
45
feedstock crops put forward, Jatropha curcas L. (henceforth referred to as Jatropha) has a long
46
and mixed history. Originally hailed as a “miracle crop” due to its predicted high oil yield under
47
marginal conditions,10–12 Jatropha has often been classified as being an ideal “pro-poor”
48
smallholder bioenergy crop.13,14 Consequently, substantial public and private investments in
49
Jatropha cultivation schemes have taken place.15,16 However, it has become apparent that the
50
requirements necessary to obtain economically viable yields have been seriously underestimated
51
and much capital has been withdrawn.14,17–21 Despite this set back, the sustainable production of
52
biofuel and other (by)products from Jatropha, albeit under more favorable growing conditions,
53
remains a concept receiving considerable political and commercial interest.16,22–25
54
As a biofuel crop there has been particular interest in the ecological impact of Jatropha
55
cultivation. The actual environmental impact of any cash crop depends on a multitude of factors,
56
which are not exclusively associated with cultivation (e.g. transport, processing, co-products and
57
consumption). Life cycle analysis is a widely accepted method used to evaluate the impact of a
58
given product (e.g. GHG emissions for biofuel) in comparison to an alternative product (e.g.
59
conventional fuel). Numerous life cycle analyses have been performed for Jatropha biofuel. In
60
many cases, however, land use change (LUC) was not included in the impact assessments.26
61
However, other analyses find that LUC can be a major determinant of the net carbon balance in
62
any biofuel supply chain.27,28 For policy makers interested in the climate change mitigation 3 ACS Paragon Plus Environment
Environmental Science & Technology
Page 4 of 37
63
potential of alternative fuels, it is essential to understand how specific crops affect land use and
64
vegetation cover. Other factors such as site establishment practices, use of chemical inputs, and
65
tillage are also important determinants of mitigation potential. While several studies have looked
66
at generic cases or empirical examples of actual practices on individual farms, to the best of our
67
knowledge, there are no broad analyses addressing the questions to what extent Jatropha
68
cultivation has i) replaced specific vegetation and land use types and ii) how the existing areas
69
are managed. As a result, the size and nature of cultivation-related impacts on a global scale
70
remain unknown.
71
In this study we aim to close this knowledge-gap by using data from a recent global survey of
72
commercial Jatropha projects to assess the major forms of land use change and management
73
practices currently associated with worldwide Jatropha cultivation. We used hierarchical cluster
74
analyses to identify groups of projects with similar i) prior vegetation and land use, ii)
75
establishment practices and iii) management practices as these are important variables affecting
76
the ecological and socio-economic sustainability of biofuel production. Prior vegetation, for
77
example, indicates the extent to which direct LUC creates a “carbon debt” that affects the GHG
78
benefits of the project. Establishment and management practices also affect the “carbon debt”,
79
and other issues such as human and animal health, soil degradation and water pollution.25
80
Intensive practices will probably have higher impacts on soil C whilst burning rather than
81
utilizing biomass as some type of co-product also affects carbon dynamics as well as that of other
82
nutrients.26
83
In a second step we calculated the corresponding area associated with the respective LUC,
84
establishment and management clusters. In contrast to regional, yet more detailed case studies,
85
we are therefore able to provide some insight into sustainability issues associated with land use 4 ACS Paragon Plus Environment
Page 5 of 37
Environmental Science & Technology
86
and management practices of current Jatropha cultivation on a global scale and provide country-
87
specific data from the survey in the supporting information for use by others.
88
89
Materials and Methods
90
Survey data
91
In this study we used empirical data from a global survey conducted in 2011 aimed at projects
92
associated with the production of plant oil from woody shrub species in tropical to subtropical
93
regions.29 The survey was based on a standardized questionnaire29 consisting of 67, both
94
structured and unstructured, questions. We use the term “project” throughout the document to
95
refer to Jatropha cultivation activities of a given entity within a single country.
96
In order to identify as many projects as possible, a non-probabilistic mixed-method sampling
97
approach was applied30,31 based on a combination of desktop research, expert interviews, and a
98
producer survey. In total, the global study was based on information gathered in over 180
99
interviews with experts and project representatives. Desktop research drew on a wide range of
100
sources, including academic publications, reports from civil society and international
101
organizations, producer websites, professional online networks as well as industry websites and
102
studies. In order to validate the desk research and identify projects that may have been
103
overlooked more than 80 semi-structured interviews were conducted with experts, consisting of
104
representatives from non-governmental organizations, industry associations, research institutions
105
and development organizations.
106
This approach yielded a target population of 401 projects. Triangulating desktop research, expert
107
and project interviews confirmed a frame population of 260 projects (i.e. those with a chance to 5 ACS Paragon Plus Environment
Environmental Science & Technology
Page 6 of 37
108
be selected due to the applied sampling), with the remaining 141 projects lacking valid contact
109
data. In order to avoid bias from sample drawing we applied a simple sampling method32 with all
110
projects from the frame population contacted for interviews between May and September 2011.
111
The questionnaire was sent out to the interviewees beforehand to give an overview of the survey
112
content. The majority of interviews with the 154 respondents (59%) were carried out by means of
113
computer-assisted telephone interviewing using an active PDF questionnaire whilst 10%
114
performed self-interviewing by sending back the filled-in questionnaire via e-mail. Interviewees
115
differed in terms of their professional status within the given project, with the majority (58)
116
occupying senior management positions (for a more detailed description see Tab. S1).
117
Despite the extensive research efforts combined with a mixed method sampling strategy,
118
systematic biases are possible. For example, specific groups of projects might be
119
underrepresented, e.g. “in-house projects” where oil-bearing woody shrubs are grown as part of
120
other business activities or are mixed with other types of feedstock production and are thus not
121
reported separately. In addition, some project owners or managers might prefer to fully develop
122
their projects and achieve reliable agronomic and economic results before exposing themselves to
123
the market and critical discussions which followed the first Jatropha hype.
124
In addition to sampling errors invalid answers and nonresponse must be considered.32 The
125
motivation for providing invalid answers can differ, e.g. if a respondent is not allowed to or does
126
not want to disclose figures such as acreage and investments or if social desirability leads to
127
strategic answers. If answers are expected to be systematically biased in one direction only (e.g.
128
exaggeration of social and environmental commitment), interpretations have to be made with
129
care. This might have been the case for certain issues addressed in this study, e.g. establishment
130
on pristine forest or on previously used (as opposed to unused) land. 6 ACS Paragon Plus Environment
Page 7 of 37
Environmental Science & Technology
131
As the main objective of this study was to assess the land use implications of operational Jatropha
132
projects we removed eight projects that did not produce Jatropha as a main crop. In addition, we
133
also removed those that were not fully operational at the time, were dedicated primarily to
134
research and development, as well as those with an unclear status or incomplete interviews. This
135
led to a final dataset of 106 fully operational Jatropha projects (Tab. S2) which were associated
136
with 84 parent companies, 33 of which stated that they were active in multiple countries.
137
Nonresponse errors affect the survey quality if they are caused by systematic biases, i.e. if the
138
nonresponding units or items differ from the sample in specific, systematic ways.33 The projects
139
from our final dataset were fairly equally distributed, with 43%, 32% and 31% situated in Africa,
140
Asia and Latin America, respectively (Fig. 1). This distribution pattern was very similar to that of
141
the complete frame population indicating there was no evidence of unit nonresponse error; at
142
least in terms of location, which was the only variable for which such a comparison was possible.
143
Due to the fact that our analysis is on a global level (i.e. we do not analyze subsamples of the
144
dataset, e.g. continents or different cultivation models) systematic item non-response errors are
145
irrelevant.
146
The 106 fully operational Jatropha projects included 32 projects that also cultivated additional
147
species, albeit to a lesser extent (secondary species in a spatially separate location), as well as 48
148
projects that grew Jatropha in combination with other crops (mixed cropping). Fourteen
149
additional projects stated that they grew Jatropha in a silvo-pastoral system. Jatropha plants were
150
mainly cultivated in rows only (n=63), six projects grew Jatropha purely as field boundary crops
151
(hedges) and 21 projects stated that they cultivated Jatropha both in hedges and rows.
152
7 ACS Paragon Plus Environment
Environmental Science & Technology
Page 8 of 37
153
Analyses
154
For the questions relevant to our study multiple responses were possible, which made it
155
impossible to calculate areas associated with specific answers. We classified projects into
156
subgroups by means of cluster analysis to overcome this problem. This is a commonly used
157
multivariate data exploration technique, helping to reveal the characteristics of structures or
158
patterns present in the data.34 We used R version 3.1.135 to conduct an agglomerative hierarchical
159
cluster analysis based on Euclidean distances and Ward’s method (package ‘cluster’, ‘function
160
‘agnes’)36 with the intent of minimizing within-group variation and maximizing dissimilarities
161
between groups.37 This method uses a bottom-up approach by initially clustering single elements
162
(in this case projects) based on the minimum variance criterion and subsequently clusters the
163
prior aggregates until one cluster remains. The variables used for cluster analyses were the
164
answers provided by interviewees in the check boxes of the relevant questions, with a checked
165
box referring to a “yes” and a blank box to a “no”. For means of analyses these answers were
166
encoded as a one and a zero, respectively.
167
We chose three categories: i) prior vegetation/ prior land use, ii) site establishment and iii) site
168
management for individual cluster analyses as these generally have the most impact on
169
sustainability issues directly linked to Jatropha cultivation and also because adequate data was
170
available. Prior vegetation refers to the type of vegetation prior to establishment of the Jatropha
171
plants whilst the term “prior land use” is independent of vegetation type and simply refers to how
172
the land was used prior to establishment of the Jatropha plants. In contrast, site establishment and
173
management practices both refer to processes directly related to Jatropha cultivation with site
174
establishment referring to initial “once off” actions required for establishing the site whilst
175
management practices such as fertilization are continuous processes occurring at a rate 8 ACS Paragon Plus Environment
Page 9 of 37
Environmental Science & Technology
176
determined by the field manager or farmer. We would like to note that initial cluster analyses
177
with all possible combinations of these categories did not lead to meaningful results, so we
178
performed separate cluster analyses for each of the aforementioned categories.
179
The relevant questions and response options for the three categories are provided in the
180
supporting information (Fig. S1).
181
We applied hierarchical cluster analysis as it is commonly used across a wide range of
182
disciplines, well understood and readily interpretable.38 The agglomerative coefficient measures
183
the clustering structure of the dataset; whereby “a value close to 1 indicates that a clear
184
structuring has been found, while a value of 0 indicates that a clear structure is missing and the
185
data represents one big cluster”.39 The results of the cluster analyses are presented as
186
dendrograms, with “height” on the y-axis indicating the level of similarity with which any two
187
clusters were joined.
188
As the identification of relevant clusters is subjective, the ability of a specific answer to define a
189
given cluster was assessed by means of indicator species analysis,40 using the ‘labdsv’ package
190
(function ‘indval’).41 In this analysis an indicator value of a variable is calculated as the product
191
of the relative frequency (“fidelity”) and relative abundance (“exclusivity”) of a given variable
192
within a cluster. Values range from 0 to 1, with the latter indicating that the variable (in this case
193
answer) of interest is a perfect indicator of the cluster (i.e. all projects within the cluster are
194
associated with the indicator variable which is exclusive to the cluster). The significance of the
195
indicator value was determined using the probability of obtaining as high an indicator value as
196
observed over 1000 iterations.41 As the original application of the indicator species analysis and
197
its description in the ‘labdsv’ package are associated with species assemblages in ecology40,41 we
198
present the formulae for calculating the relative frequencies and abundancies within the ‘labdsv’ 9 ACS Paragon Plus Environment
Environmental Science & Technology
Page 10 of 37
199
package using the terms from our analysis (equation 1 and 2, respectively). For better comparison
200
we include the original terms used in the ‘labdsv’ in italics within parenthesis. Note that in
201
contrast to the original application for species assemblages the abundance of an answer within a
202
project (compared to a species within a sample) can only take on the values zero or one.
203
For a cluster c in set K where:
204
, presence/absence (1/0) of answer (species) i in project (sample) j
205
, abundance of answer “yes” (species) i in project (sample) j
206
number of projects (samples) in cluster c
207
Equation 1:
relative frequency: , =
∑ ∈ ,
(∑ ∈ % , )/ (*+((∑ ∈( % , )/ ( )
relative abundance: #, = ∑)
208
Equation 2:
209
With regards to prior vegetation, land use, establishment and management practices we found no
210
indication of significant differences between those projects that cultivated Jatropha in boundary
211
(hedges) or in mixed cropping systems and those that did not (Chi2 test). Therefore, we did not
212
distinguish between such systems in this study.
213
As questions allowed for multiple answers we were only able to attribute the corresponding area
214
of a project to a given answer (e.g. “degraded land”) when a single answer was provided. In order
215
to provide some insight into the extent of area associated with the different land use types and
216
management practices at different geopolitical scales we calculated an “area range” by summing
217
the minimum and maximum area for each project over countries and continents, which were
218
calculated as follows. For those cases in which more than one answer was given, we set the 10 ACS Paragon Plus Environment
Page 11 of 37
Environmental Science & Technology
219
minimum area per answer to 1 ha, assuming this to be a conservative estimate of the average
220
minimum size of a cultivated plot. Accordingly, we calculated the maximum area by subtracting
221
1 ha from the total area per project. For projects were only one answer was given minimum and
222
maximum values were identical. While the actual area lies somewhere within the range provided,
223
we considered it important to present these estimates and add the data to the supporting
224
information for possible further use by others (figures S3-S5 and tables S4-S8).
225 226
Results
227
Prior Vegetation and Land Use
228
In total, information from 95 projects on both the prior vegetation and land use type was available
229
(90% response rate, Tab. S3). The four main clusters identified were generally characterized by
230
their prior land use and to a lesser extent prior vegetation type. Accordingly, the clusters were
231
named for the prior land use which was most representative, i.e. “Grazing”, “Crops”, “Unused”,
232
and “Logged” (Tab. 1; Fig. 2a). The latter encompassed only nine projects two of which belonged
233
to one large parent company active in Asia. In this context, the term “Logged” mainly refers to
234
previously logged land with only two projects reporting an actual removal of trees as a direct
235
consequence of Jatropha establishment. Therefore, the Logged cluster can best be described as
236
small group of projects primarily located on degraded land that was cleared of vegetation as a
237
result of logging activities unrelated to Jatropha cultivation.
238
“Degraded land” was the most frequent prior vegetation type occurring in 66% of all projects. It
239
was not a significant cluster indicator, however, as it was fairly equally distributed across clusters
240
with the exception of the Crops cluster where it was less common but nonetheless still occurred
241
in 50% of projects. In contrast, the most frequently named prior land use “unused” (34 projects) 11 ACS Paragon Plus Environment
Environmental Science & Technology
Page 12 of 37
242
was a perfect indicator for the identically named cluster. The Unused cluster can therefore best be
243
described as large group of projects exclusively located on formerly unused land, often with
244
degraded savannah or shrubland (defined as an area where shrubs without a definite crown are
245
the dominant vegetation, see Fig. S1) as prior vegetation.
246
The prior land use variable husbandry/pasture best characterized the Grazing cluster. The
247
remaining projects with husbandry/pasture as a prior land use were found solely within the Crops
248
cluster. Unsurprisingly, there were strong positive correlations between food crops, non-food
249
crops and farmland (Chi2, p
