1 - Rizky Rahmadhani
1 - Rizky Rahmadhani
1 - Rizky Rahmadhani
Abstract
Biodiversity hotspots like tropical lowland rainforests in Sumatra are threatened by the agricultural expansion, which
increases the deforestation rate in Indonesia, which is highest worldwide. Main land-use change drivers in Indonesia include
the production of rubber and palm oil, both of which lead to a high forest conversion rate. In the remaining and degraded
forest patches, species diversity has declined and species composition has been altered. Effects of habitat fragmentation and
land-use change on genetic structure were frequently investigated at the species level and compared across plant species, but
not for plant communities. In addition, the effect of land-use change on the genetic structure of plants has not yet been
investigated. Using Amplified Fragment Length Polymorphism marker, the genetic diversity of 112 dominant plant species
was assessed in four different land-use systems in Sumatra: old growth tropical lowland rainforest, jungle rubber, rubber
plantation, and oil palm plantation. The four systems were investigated in two regions with four replicates, respectively.
Because of different species compositions, characterized by different life history traits, forest and jungle rubber plots showed
the highest diversity level, while oil palm and rubber plantations showed the lower diversity levels. The two intensively
managed plantation systems showed similar genetic diversity levels as the tree dominated systems but are dominated by
mainly alien species. This indicates that oil palm and rubber plantations could not be identified as habitats of conservational
value. The newly introduced collection and analysis approach presents a universally applicable method to investigate different
ecosystems in their plant genetic diversity to support the identification of habitats with high conservational value.
Keywords
plant genetic diversity, land-use change, AFLP, oil palm, rubber, Hevea brasiliensis, Elaeis guineensis
Introduction
Threats to Biodiversity
Globally, tropical forests harbor two thirds of terrestrial 1
Department of Forest Genetics and Forest Tree Breeding, Georg-August
species, while covering only approximately 5% of the University of G€ ottingen, Germany
2
global surface (Gardner et al., 2009). Most tropical Centre for Plant Conservation, Bogor Botanic Gardens, Indonesian
regions are, on one side, categorized as biodiversity hot- Institute of Science, Indonesia
3
Department of Silviculture, Faculty of Forestry, Bogor Agricultural
spots and on the other side, have above-average human University, Indonesia
population growth rate (Cincotta, Wisnewski, & 4
Forestry Faculty, Jambi University, Indonesia
5
Engelman, 2000). Biodiversity hotspots are character- University of Kassel, Germany
ized by a high number of endemic species, high species Received 2 August 2018; Revised 30 September 2018; Accepted 28
richness, and increased habitat loss (Myers, 1988). October 2018
Causes of deforestation are the expansion of agricultural Corresponding Author:
Natalie Breidenbach, Department of Forest Genetics and Forest Tree
land-use, logging, and mining activities at high rates for Breeding, Georg-August University of G€ ottingen, Büsgenweg 2, 37077
the last three decades (Abood, Lee, Burivalova, Garcia- G€
ottingen, Germany.
Ulloa, & Koh, 2015; Food and Agriculture Organization Email: [email protected]
Creative Commons Non Commercial CC BY-NC: This article is distributed under the terms of the Creative Commons Attribution-
NonCommercial 4.0 License (http://www.creativecommons.org/licenses/by-nc/4.0/) which permits non-commercial use, reproduction and dis-
tribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.
sagepub.com/en-us/nam/open-access-at-sage).
2 Tropical Conservation Science
[FAO], 2010; Lambin, Geist, & Lepers, 2003). (Honnay & Jacquemyn, 2007; Mona, Ray, Arenas, &
Consequently, the most diverse ecosystems are under Excoffier, 2014; Vranckx, Jacquemyn, Muys, &
increasing danger of biodiversity degradation and spe- Honnay, 2011).
cies extinction, because of human expansion (Brooks In theory, consequences of habitat loss at the intra-
et al., 2006; Cincotta et al., 2000; Sloan, Jenkins, specific diversity level are the loss of genetic diversity and
Joppa, Gaveau, & Laurance, 2014). increased differentiation of populations, caused by
Sumatra belongs to the Indonesian biodiversity hot- genetic drift, inbreeding and isolation by distance, and
spot Sunda-land (Myers, Mittermeier, Mittermeier, Da increased differentiation of populations, but not all
Fonseca, & Kent, 2000). In Indonesia, the area of annual empirical studies confirm this (A. T. Kramer, Ison,
forest cover loss increased in 11 years from approximate- Ashley, & Howe, 2008; Lesser, Parchman, & Jackson,
ly 0.22 Mha in 2001 to 0.84 Mha in 2012 (Margono, 2013; Sampson et al., 2014). The different consequences
Potapov, Turubanova, Stolle, & Hansen, 2014). Since of population fragmentation for species are caused by
the 1970s, the two nonnative crop species oil palm different life history traits such as, gene flow, dispersal
(Elaeis guineensis Jacq.) and rubber, Hevea brasiliensis strategy, mating system, degree of isolation, and tree
(Willd. ex A. Juss.) Müll. Arg., became the major crop density of populations (Bacles & Jump, 2011; Breed
species planted in monoculture (both species) and agro- et al., 2015; Ganzhorn, Perez-Sweeney, Thomas,
forest systems (rubber) in Sumatra (Noordwijk, Gaiotto, & Lewis, 2015; Jacquemyn, De Meester,
Suyamto, Lusiana, Ekadinata, & Hairiah, 2008; Jongejans, & Honnay, 2012; Kashimshetty, Pelikan, &
Villamor, Pontius, & Noordwijk, 2013). E. guineensis is Rogstad, 2015; Kettle, 2014; Sebbenn et al., 2008). Thus,
currently considered as the biggest and main threat to it is not advisable to develop conservation strategies of
biodiversity in Southeast Asia (Wilcove & Koh, 2010). genetic resources for a plant community based on results
In 25 years, the oil palm production area in Indonesia of one or few species only. A universal approach is
reached approximately 12 million hectares in 2017 needed, which is able to compare entire plant communi-
(Hendaryati et al., 2017) and since 2008, the country ties concerning their genetic variability.
became the largest producer of palm oil worldwide
(FAO, 2014). H. brasiliensis was introduced in Genetics in Conservation Management
Indonesia around 1900. Rubber was primarily produced
The conservation of natural ecosystems depends on
in an agroforestry system, called “jungle rubber,” which
human intervention and management (Sloan et al.,
is an integrated complex agricultural system within sec-
2014). Different definitions and goals of tropical forest
ondary forest vegetation and can be seen as more sus-
management techniques can lead to contradictory results
tainable rubber production (Gouyon, Foresta, &
(Lynam, De Jong, Sheil, Kusumanto, & Evans, 2007).
Levang, 1993; Michon & Foresta, 1995). Between 1993
Furthermore, economic and social factors, such as high
and 2005, rubber production was shifted from agroforest
human population growth, corruption, agricultural
to monoculture (Villamor et al., 2013). The rubber pro-
expansion, and the lack of experience in nature conser-
ducing area increased from 1.9 Mha in 1990 to 3.6 Mha
vation strategies, increase the difficulty to manage trop-
in 2013 and has remained at this level until 2017 (FAO,
ical ecosystems in a sustainable manner and to assess
2014; Hendaryati et al., 2017).
their conservation value (Kettle, 2014; Sodhi, Koh,
Brook, & Ng, 2004; Wilcove, Giam, Edwards, Fisher,
Consequences of Land-Use Change & Koh, 2013). The growing area of secondary forests
Land-use change and the agricultural intensification in and the efforts to maintain connectivity among remain-
Indonesia have a global effect on biodiversity and ing natural habitats enhance challenges for predicting
human welfare with an impact which can only be esti- parameters to assess habitats of conservational value
mated (Foley, De Fries, Asner, Barford, & Bonan, 2005; (HCV; Roundtable on Sustainable Palm Oil
Laurance, Sayer, & Cassman, 2014; Newbold et al., [RSPO], 2013).
2015). In general, the most severe consequences, habitat Genetic information of a species can support the eval-
loss and fragmentation, result in the decline of species uation of survival probabilities and conservation
diversity and change of plant species composition requirements of the target species (Leimu, Vergeer,
(Rodrıguez-Echeverry, Echeverrıa, Oyarzun, & Angeloni, & Ouborg, 2010; Sthultz, Gehring, &
Morales, 2018; Savilaakso et al., 2014; Sodhi et al., Whitham, 2009) and help to maintain or even increase
2010), in an alteration of ecosystem processes its viability in the context of landscape management
(Laliberté & Tylianakis, 2012), and an increased propor- (Bozzano et al., 2014; Li & Jin, 2007; Thomas et al.,
tion of invasive species (Rembold, Mangopo, 2014). Genetic diversity of plants is not only important
Tjitrosoedirdjo, & Kreft, 2017). Further, for many spe- for the survival of the plant species itself, but also influ-
cies, an alteration of genetic structures was observed ences other levels of organization and dynamics within
Breidenbach et al. 3
the ecosystem (Bailey et al., 2009; Whitham et al., 2006), different agricultural intensity are compared with old-
especially when genetic resources of dominant or key- growth tropical lowland rainforest: jungle rubber < rub-
stone species are concerned (Crawford & Rudgers, ber plantation < oil palm plantation, totaling in four
2013; Hughes, Inouye, Johnson, Underwood, & land-use systems (Figure 1). Each of these investigated
Vellend, 2008; Whitham et al., 2003). Suggestions for land-use systems had four replicates in two regions, one
sustainable management and monitoring programs of named after the national park “Bukit Dua Belas” and
particular forest areas are mainly based on data about one after the forest restoration concession “Harapan.”
species richness, particular indicator species, and ecosys- The distance between the two regions was approximately
tem processes (Bustamante et al., 2016; Corona, Chirici, 55 km. In each of the 32 (4 replicates 4 land-use sys-
McRoberts, Winter, & Barbati, 2011; Rametsteiner & tems 2 regions) 50 m 50 m plots, 10 vascular plant
Simula, 2003; Thompson et al., 2013), and have not species, dominant by biomass, were selected using
always been sufficient and adequate to maintain impor- the “Bitterlich-Method” (H. Kramer & Akca, 2008).
tant ecosystem structures and services (Porter-Bolland The total sample size was 10 (plants/species) 10 (spe-
et al., 2012; Stoll-Kleemann, 2010). Various studies cies/plot) 32 (plots) ¼ 3,200 plants, including 112 dif-
(Lande, 1988; Laurance et al., 2012; Symes, Rao, ferent species. The “Bitterlich-Method” was applied at
Mascia, & Carrasco, 2015) have emphasized the impor- 16 predefined positions in each of the plots (online
tance of investigating aspects concerning population
Appendix 1). In the plantation systems, the dominant
genetics in a conservation context but do not include
nonwoody, noncrop species were identified by the short-
genetic diversity as a parameter to identify habitats or
est distance to the same 16 positions used in the tree-
landscapes to be conserved. Hawthorne and Abu-Juam
dominated systems. If necessary, individuals were also
(1995) developed a Genetic Heat Index based on the
sampled in the area surrounding the plots (up to 300
abundance of rare or common species in a particular
m). For the noncrop species, collected individuals had
area to assess conservation priorities. This index is
not based on genetic data, and hence, it cannot consider a minimum distance of 10 m to each other to minimize
the far reaching consequences of habitat loss for the risk of clone sampling in the plantation systems. The
each species. two crop species H. brasiliensis and E. guineensis
The lack of information on genetic diversity, com- occurred in high numbers in the transformed systems,
bined with different consequences of population frag- and hence, the sampled individuals were chosen random-
mentation for various species (Kashimshetty et al., ly. Species identification was carried out with the sup-
2015) can result in inefficient management strategies to port of the EFForTS-subproject B06, the staff of Kebun
conserve plant communities. In an era of fast and vast Raya, Bogor, and Harapan. From each selected individ-
land-use change in the tropics, an applicable and effec- ual of each species, leaf material was sampled and dried
tive method for identifying biodiversity hotspots and in sealed plastic bags filled with silica gel. Herbarium
determining protected areas is necessary (Ferraro et al., specimen were collected from all individuals, excluding
2015; Lee, Garcia-Ulloa, & Koh, 2011; Reed, Sarasan, crop species, and stored at Jambi University.
Kane, Bunn, & Pence, 2011). At the genetic diversity
level, methods for the determination of hotspots in DNA Extraction and AFLP Analysis
plant communities are not established yet (Souto et al.,
DNA extraction was carried out using the Dneasy 96
2015). We introduce a universally applicable method to
Plant kit and its manufactory’s protocol (Qiagen,
investigate genetic diversity of a high number of plant
Hilden, Germany). The total genomic DNA was
species. The objective is to compare the genetic diversity
extracted out of approximately 1 cm2 dried leaf material.
of the dominant plant species community in different
The DNA was stored at 20 C.
land-use systems because of different species composi-
Amplified Fragment Length Polymorphisms (AFLPs)
tions. These results can enrich biodiversity baseline
were investigated according to the protocol of Vos et al.
data to support the identification of habitats of high
(1995) with minor modifications. All samples were ana-
conservational value (HCV; RSPO, 2013) and conserva-
tion management of threatened landscapes. lyzed with the one single enzyme/primer combination.
The restriction was carried out simultaneously with
the EcoRI/MseI primer combination and incubat-
Methods ed overnight.
For the preselective polymerase chain reaction (PCR),
Study Sites and Sample Collection the E01/M03 primer combination was used and as selec-
As a part of the interdisciplinary collaboration project tive E35/M63 (Keygene N.V. nomenclature). The PTC
“EFForTS” in Sumatra, Indonesia (Drescher et al., 200, Pelmer thermal cycler (MJ Research, Hampton,
2016), three land-use systems in Jambi Province, with USA) was used for the PCR, and the applied preselective
4 Tropical Conservation Science
Figure 1. The investigated old-growth forest (a) and the land-use systems jungle rubber (b), rubber plantation (c), and oil palm plantation
(d) in Jambi Province, Sumatra, Indonesia.
and selective PCR-program protocols were according to study (minimum 10, maximum 160) and the high varia-
Kuchma (2010). tion in number of AFLP-loci between species (minimum
The in-water-diluted selective PCR product was sep- 37, maximum 212). These differences restrict the use of
arated using the ABI genetic analyzer (3130, ABI common genetic diversity indices (Bonin, Ehrich, &
PRISM, Waltham, USA), and the size standard Manel, 2007). For example, the Shannon Index
GENSCAN 500 ROX was added to the solution. (Shannon & Weaver, 1949) can be biased because of
Fragment determination was carried out with the the different number of samples (Goodmann, 1975).
program GeneMapper 4.1. (Applied Biosystems, In this study, however, the Shannon Index was applied
Waltham, USA). to the varying number of AFLP fragments per species.
The reproducibility of the AFLP profiles was assessed For that reason, genetic diversity was calculated using
with two samples of each species, which were repeated two different approaches. For the first approach to cal-
from DNA extraction onward. Species with only 10 indi- culate genetic diversity, the Shannon’s information
viduals, a second repetition for all samples was con- Index (I; Shannon & Weaver, 1949) was used because
ducted from the restriction step onward and only of its commonness and comparability with other studies.
fragments occurring in both repetitions were considered. Genetic diversity (I) was calculated based on 10 individ-
Only unambiguous and reproducible fragments were uals per species in each plot respectively with the pro-
used for the analysis. The results were transformed gram PopGen1.32 (Yeh & Boyle, 1997). Considering the
into a fragment present–absence (1–0) matrix. limitations of the Shannon Index, a second approach
was carried out. For this second approach, each
fragment of all sampled species is considered as a unit
Data Analysis of genetic diversity. As in phylogenetic analyses
All analyses were based on the 1–0 matrices for each of (Robinson & Harris, 1999), each locus was assumed to
the 104 successfully genotyped species. For the following be an independent unit. All scored loci of all 104 species
analyses, two aspects were considered: the high variation in all 32 plots are presenting a pool of fragments in
in number of individuals per species collected in the which a single AFLP fragment of a species can be
Breidenbach et al. 5
present. The 100 individuals per plot were combined into conducted by using the function “betadisper” in the R-
10 fragment pools per plot. Every plot was presented by package “vegan” (Oksanen et al., 2015). The graph was
10 rows with each successively 10 individuals with their built with the R-package “ggplot2” (Wickham, 2009).
individual 1–0 order. The first row contained successive-
ly the first individuals, each with the respective 1–0 Genetic Diversity and Differentiation Levels Within
AFLP score, of every species collected in this plot and Among Land-Use Systems
(online Appendix 2). Each fragment was placed in one
column. The second row contained all second individu- To assess genetic diversity with increasing population
als, each with the respective 1–0 AFLP score, of every size of the dominant species and to test the genetic diver-
species collected in this plot. Hence, the 1–0 AFLP fin- sity dependency on the land-use system, different spatial
gerprint of the second individual of the same species was scales were established. The different spatial scales are
in the same columns as the first individual. This was presented by the three levels: the lowest, a-level, corre-
continued for all 10 individuals and 10 species for all sponds to the diversity within each plot; the b-level
plots. In the following, these rows are called fragment to diversity within each land-use system; and the
pools. If the species occurred again in a different plot, highest, c-level, to the diversity within each region, that
the 1–0 matrix for these individuals was accordingly is, the diversity within the 16 plots in each region (online
placed into the same columns. This concept was fol- Appendix 1).
lowed for all species and plots. Online Appendix 2 illus- For the fragment pool approach, the a-level differen-
trates an example of the fragment pool approach with tiation was calculated by taking the mean of the 10 pair-
seven species and four plots. wise genetic fragment pool distance values within each
The pairwise Morisita–Horn dissimilarity index plot. The b-diversity level was calculated by taking the
(Horn, 1966) based on the fragment pool approach mean values of the 40 pairwise genetic fragment pools
allows to compare plots and land-use systems genetically within each land-use system, for example, of the pairwise
while simultaneously considering differences in species genetic distance values from all 40 forest fragment pools
composition. To compare dissimilarities among plots in the region Harapan. The c-diversity level was based
and land-use systems only because of the genetic struc- on the mean values of the 160 pairwise genetic fragment
ture of the species, the species effect was accounted for pool values within each region. The results were 10 mean
by the following procedure. A second input file was built values per plot for the fragment pool for all three diver-
where all individuals were considered to be clones, that sity levels, respectively. The genetic diversity is repre-
is, all fragments of the occurring species in each plot sented by the Shannon Index (Lewontin, 1972;
were present (online Appendix 2a). Based on these Shannon & Weaver, 1949) among individuals of each
clone fragment pools, the Morisita–Horn pairwise dis- species in one plot, a-level; among individuals within
tance matrix was calculated. Herewith, the resulting dis- each land-use system, b-level; and within each region,
tances between the clone fragment pools are only c-level. The a-level calculation is based on 10 individuals,
because of the species differences and the genetic diver- b-level on 10 to maximum 40 individuals, and c-level on
sity is zero. The difference between the two pairwise dis- 10 to maximum 160 individuals depending on the
tance matrices, the fragment pool matrix and the clone number of plots the species was dominant.
fragment pool matrix, corresponds to the pairwise genet- The data sets, a-, b-, and c-diversity levels for genetic
ic fragment pool distance matrix. This pairwise genetic diversity and differentiation were checked for normal dis-
fragment distance matrix was used for the calculations tribution using the Kolmogorov–Smirnov test (Stephens,
among fragment pools. For this calculation, the 1979) for continuous data in STATISTICA version 12
Morisita–Horn dissimilarity index was used, which cal-
(StatSoft Inc., Tulsa, USA). Results were graphed using
culates the number of shared fragments to the overall
R-package “reshape2” (Wickham, 2007). Differences
number of fragments in the two compared samples
among the land-use systems in genetic diversity and dif-
(Bonin et al., 2007). The calculation was carried out
ferentiation at the a-, b-, and c-diversity levels were
using the “vegdist” function of the R-package vegan
assessed using a generalized mixed fixed effect model in
(Oksanen, Blanchet, & Kindt, 2015).
the R-package nlme (Pinheiro, Bates, Debrov, Sarkar, &
R Core Team, 2015). The model of the best fit was chosen
Heterogeneity of the Analyzed Plots and based on the Akaike Information Criterion: the land-use
Land-Use Systems system as the fixed variable and the plots nested in a
The heterogeneity of the land-use systems was visualized region as the random effects. The results of the mixed
using the mean value of dispersion of each plot in a effect models were generalized for multivariate compari-
Principal Component Analysis (Gower, 1966) based on sons with the function “glht” of the R-package
the fragment pool distance matrix. The calculation was “multcomp” (Hothorn, Bretz, & Westfall, 2008).
6 Tropical Conservation Science
Figure 2. PCoA based on genetic and species dissimilarity of all 32 plots. BF ¼ Bukit Duabelas forest, BJ ¼ Bukit Dua Belas jungle rubber;
BO ¼ Bukit Dua Belas oil palm plantation; BR ¼ Bukit Dua Belas rubber plantation; HF ¼ Harapan forest; HJ ¼ Harapan jungle rubber;
HO ¼ Harapan oil palm plantation; HR ¼ Harapan rubber plantation; PCoA ¼ Principal Component Analysis.
Breidenbach et al. 7
Figure 3. Mean genetic differentiation (a) and diversity (b) per land-use system F (forest), J (jungle rubber), O (oil palm plantation), and R
(rubber plantation) for the three spatial levels a (within plot), b (within land-use system), and c (within region). Significant differences
indicated by letters (p < .01). Values for each species and fragment pool are shown in online Appendices 4 and 5.
four land-use systems (Figure 3). The results at the et al., 2016; Rembold et al., 2017). Current results indi-
a-diversity and b- diversity levels revealed two groups cate that land-use change per se does not have an effect
with significant difference: forest and jungle rubber on genetic diversity of the analyzed dominant plant
with high mean genetic diversity and oil palm and species, but has an impact on the particular species
rubber plantations with low mean genetic diversity. composition with their respective genetic characteristics.
Species of rubber plantations showed slightly higher
mean diversities than oil palm plantation species. Heterogeneity of the Analyzed Plots and
Genetic diversity differences among the four land-use Land-Use Systems
systems at c-diversity level were low and no significant The uniqueness of the dominant species composition in
differences could be detected. Mean value for jungle each plot and land-use system was only partly con-
rubber is slightly higher than the forest mean values. firmed. The two plantation systems were very similar
As for genetic differentiation, the mean genetic diversity concerning dominant species composition and showed
for the forest species did not change from b- to c-levels. low genetic differentiation between plots and regions.
Variance within each land-use systems was high and As expected, jungle rubber plots present an intermediate
increased with increasing spatial scale. system between forest and rubber plantations, having
trees as the dominant life form and sharing with the
Discussion rubber plantation at least one species, H. brasiliensis.
Dominant species of the plantations systems possibly
The four investigated land-use systems reflect different occur in jungle rubber and forest but do not dominate
agricultural intensities with specific dominant plant spe- the system in biomass. Differences and variability
cies composition. On the level of genetic diversity, obser- between plots of each plantation system and jungle
vations concerning the consequences of land-use change rubber occurred mainly because of different manage-
were not as clear as previously observed species diversity ment strategies (Drescher et al., 2016; Rembold et al.,
studies conducted on the same research sites (Drescher 2017). In general, the structure and species composition
8 Tropical Conservation Science
depend among others on the degree of disturbance in genetic diversity using the Shannon Index as well, but
jungle rubber (Laumonier, 1997; Schroth, Harvey, & because of the high variance within each land-use
Vincent, 2004) and as well in forest (Burivalova, system, the effect of these results was not visible.
Şekercio
glu, & Koh, 2014). Compared with the forest Differences between b- and c-levels were based on an
with a closed canopy, the higher temperature in the even more reduced number of species. The land-use sys-
more open jungle rubber and plantation systems alters tems of oil palm and rubber plantation share seven spe-
the microclimatic conditions, which increases the effect cies. Forest and jungle rubber share three species,
of land-use change concerning biodiversity and ecosys- Macaranga bancana (Miq.) Müll.Arg., Endospermum
tem processes (Hardwick et al., 2015; Drescher et al., malayanum (Pax & K.Hoffm) Chatterjee, and Parkia
2016). Further, landscape dynamics, ecosystem hetero- speciosa Hassk., with each other. M. bancana was also
geneity, and the niche size is influencing the community dominant in two rubber plantations in the Harapan
similarity regarding their phylogenetic and intraspecific region, P. speciosa in one rubber plantation in
variance (Gascuel, Ferrière, Aguilée, & Lambert, 2015; Harapan. E. malayanum only occurred in the Bukit
Parks & Beiko, 2012; Violle et al., 2012). Dua Belas region. Together, a maximum of two plots
of different land-use systems per species were shared.
Genetic Diversity and Differentiation Levels Within Hence, for most tree species, the differentiation or diver-
and Among Land-Use Systems sity results did not change from the b-level to the c-level
(online Appendices 4 and 5). In contrast, the seven spe-
Three spatial scales (within plot, within land-use system, cies shared by the plantations occurred in the most of the
within region) were investigated to assess the mean 16 plots. Hence, more populations were considered in
genetic differentiation and diversity of the dominant the analyses and the changes in results were higher.
plant species with different potential population sizes Greater changes from a- to b-/c-level, using both
and the dependency on the land-use systems. The tree- approaches, in the plantation systems compared with
dominated systems were expected to have high local the tree-dominated systems can be explained by the
(a-level) genetic diversity and lower differentiation on a higher similarity in the species composition and the
larger scale (c-level), while the plantation systems dom- differences in gene flow between nonwoody and
inated by vegetative growing species may show the woody species (Austerlitz, Mariette, Machon, Gouyon,
opposite pattern. Analyses of genetic diversity and & Godelle, 2000; J. Hamrick, Godt, & Sherman-
genetic differentiation show results that vary from the Broyles, 1992).
aforementioned hypothesis. For the fragment pool The influence of each index is intensified by the
approach, the observed decreasing level of differentia- unique number of AFLP loci, abundance, and particular
tion with increasing spatial size was not expected. The genetic differentiation of each species. Using the frag-
mean genetic diversity values for all land-use systems, ment pool approach, the total number of loci of all spe-
calculated using the Shannon Index, increased only cies occurring in the respective plot/land-use system/
little from a- to c-diversity levels. region was included in every distance calculation.
The unexpected variation in results of mean genetic Hence, with increasing spatial scale, the number of loci
differentiation and diversity are because of the depen- is increased while the weight of the genetic differentia-
dency on the dominant species composition with their tion among individuals of the species, dominant only in
abundance within and among land-use systems and on one plot, decreased. Shannon Index values were calcu-
the nature of the two indices (White, 1986). The influ- lated for each species separately and only changed for
ence of the species composition is mainly because of the the species occurring in more than one plot/land-use
very different abundance of each species. Differences system/region, which increased the variance within
between a- and b-levels are based on the species which each land-use system with increasing spatial scale. The
occurred in more than one plot within each land-use main disadvantage when comparing mean values per
system and region. For the forest, 38% (13 species of species is the high variance within the land-use system
34) of the tree species were dominant in more than one and the disadvantage of the fragment pool is the high
plot, in jungle rubber 24% (8 species of 33), and in the effect of extreme values, which increases the high effect
two plantation systems it is 45% (19 species of 42). Eight of potential outliers and the decreasing influence of each
of the 13 forest species showed a decreased genetic dif- locus with increasing scale. Because of the increased
ferentiation within the land-use system in comparison number of loci considered and increasing variance with
with the mean differentiation within plots, which results increasing spatial scale, the three diversity levels are only
in a decrease in of pairwise distance among each frag- comparable in each level but are less comparable
ment pool within the forest system. Three of these spe- among levels.
cies (Porterandia anisophylla, Gironniera nervosa, and Further, results may be influenced by the sampling
Baccaurea sp. III; online Appendix 4) showed decreased method, the chosen genetic marker and the calculation
Breidenbach et al. 9
methods for the estimated diversity parameters. 2017) and do not show any signs of genetic diversity
Dominant species are expected to have highest influence loss. All invasive species have already been introduced
on the ecosystem and may represent them most (Avolio, in Indonesia decades before this study and because of
Chang, & Smith, 2011; Grime, 2001). Using the their rapid growth and possible multiple introductions,
“Bitterlich Method” to choose the dominant tree species which makes the possible genetic consequences of a bot-
in every plot assures the randomization and objectivity tleneck undetectable. But a comparison with native pop-
of the selection. Selecting species with only 10 individu- ulations would be necessary (DeWalt & Hamrick, 2004;
als in or close to the 50 50 m plot means neglecting Dlugosch & Parker, 2008). Further, the differentiation
large trees with low densities. Outcrossing species which may be as high as the tree dominated systems because of
recently declined in their population size, for example, the size of the population/plots. For herbaceous species,
because of deforestation, are assumed to be most threat- one plot can be considered as one population, that is, at
ened by genetic consequences of forest fragmentation b-level, the differentiation among plots are higher than
(Aguilar, Quesada, Ashworth, Herrerias-Diego, & for tree species which can have a larger geographic pop-
Lobo, 2008). Albeit, a larger investigation area could ulation size, that is, several plots in one region. Despite
not be defined because of the small jungle rubber patches the similar genetic diversity levels in tree dominated and
and is still remaining in the area. plantation systems, the dominance of herbaceous spe-
As expected, the mean genetic diversity depends on cies, accompanying the crop species, and the high abun-
the species dominating the respective land-use system dance of nonnative colonizing species in the latter,
but not on the land-use system itself. The comparatively exclude the plantations from preserving forest genetic
low values for forest and jungle rubber and their small diversity and its resources. Therefore, the plantation sys-
differences to plantation systems may indicate that the tems are not considered as habitats of conservation-
analyzed species in forest and jungle rubber might have al value.
experienced limitations in gene flow and genetic drift, as
a consequence of population fragmentation (Bacles & Molecular Marker and Statistical Analyses
Jump, 2011; Vranckx et al., 2011; Young, Boyle, &
Brown, 1996). Differences in mean genetic diversity AFLP markers were preferred to other molecular
and differentiation between single plots and the regions markers for the following reasons: In this study, a high
could be detected, although not significantly (data not number of species was expected and prior knowledge
shown). Jambi Province experienced high forest trans- about the species’ DNA sequence is for the anonymous
formation rates since the 1970s, resulting in a land AFLP method not necessary. Fragments of the same size
cover of more than 50% of crop producing land-use can originate from different regions of the DNA, which
systems. Accessible forest plots for this study were clas- is a major advantage of the method, especially for poly-
sified as “primary, degraded” (Drescher et al., 2016). The ploid species (Després, Gielly, Redoutet, & Taberlet,
fragment pool approach suggests for the forest land-use 2003; Goldman et al., 2004) and tropical species for
system differences between particular plots and regions. which taxonomic identification are often ambiguous
The forest plots in the region Harapan seem more homo- (Kremer et al., 2005; Mace, Gebhardt, & Lester, 1999).
geneous (Figure 1). Despite low genetic differentiation, The dominant nature of the AFLP marker limited the
these four forest plots might be of higher conservational availability of analytical methods that could be used in
value than the forest plots in Bukit Dua Belas, because this study. Three further aspects needed to be considered
the individuals of the four plots seem to be one, more for the analysis, which led to the use of AFLPs as a
continuous and larger, population. Laumonier et al. molecular marker and the Morisita–Horn and the
(2010) identified Harapan as a forest with high priority Shannon Indices as the diversity measures. First,
for conservation because of its size and comparable low AFLP fragments of different species could not be com-
disturbance. The forest in Bukit Dua Belas region seems pared as the origins of fragments of the same size are
more fragmented, individuals in BF1 and BF2 showed unknown. Second, the different weight per species within
higher differentiation to all other individuals of shared the analysis because of the varying number of collected
species than all other plots (online Appendix 5). The individuals (Mba & Tohme, 2005; Meudt & Clarke,
increased differentiation might indicate that the two 2007). Third, every species has its unique number of
plots were experiencing limitations in gene flow (Hahn, fragments, which varied highly among the sampled spe-
Kettle, Ghazoul, Hennig, & Pluess, 2013; Wang & cies (Linton, Davies, & Wrona, 1981; Wolda, 1981).
Bradburd, 2014). In the plantations, dominant species Recent studies about plant genetic structure have not
from the three life forms herb, grass, and fern are used the Morisita–Horn dissimilarity to analyze genetic
mainly alien or colonizing species and hence are adapted diversity. Hence, for a better comparison with previous
to disturbance (DeWalt & Hamrick, 2004; Ootsuki, studies, the Shannon Index (Shannon & Weaver, 1949)
Sato, Nakato, & Murakami, 2012; Rembold et al., was also used as a common genetic diversity measure.
10 Tropical Conservation Science
In this project, several species with a high variation in value but decreased crop productions have an increased
fragment number were compared, which can lead to a probability to be converted to monocultures or rejuve-
bias using the Shannon Information Index. Further, the nated (Rembold et al., 2017; Schroth et al., 2004).
Morisita–Horn index is widely used as a b-diversity Furthermore, the presence of agroforestry in a region
index and the Shannon Index as an a-diversity index in does not necessarily protect natural stands from exploi-
ecology. Both were compared at the three different scales tation (Dawson et al., 2013). Schemes like Payment for
during this study. In ecology, “differentiation” and Ecosystem Services, which are considered to increase the
“diversity” are used as synonyms when calculating protection of forests (Lee et al., 2011), could also be a
b-diversity (Koleff, Gaston, & Lennon, 2003). possibility to protect agroforest systems of high conser-
vational value.
Implications for Conservation However, it is necessary to have a universal applicable
method to identify HCV habitats, especially in the
Despite differences in mean genetic differentiation that
highly heterogeneous and fast changing tropical ecosys-
were detected using the fragment pool approach, a gen-
eralization about each land-use system is not possible. tems. Considerations of genetic diversity of plants in
This may be because of very different genetic diversity conservation management plans increases the capability
results within each plot and land-use system in this proj- to conserve global biodiversity of similar natural
ect, which confirm the high dependency of genetic diver- ecosystems, not only in Indonesia, particularly with
sity and structure on very different life forms, regard to future challenges caused by forest loss and
reproduction systems, dispersal strategies, dispersal climate change.
ranges and population densities, among species (Ewers
& Didham, 2006; J. Hamrick et al., 1992; J. L. Hamrick, Acknowledgments
Linhart, & Mitton, 1979; Sebbenn et al., 2008). For con- The authors thank the landowners of the plots and the staff of
servation purposes, it would be best to investigate each Taman Nasional Bukit Dua belas and Harapan Rainforest.
species in detail, to be able to consider their character- The authors thank the herbarium staff of Harapan
istics. That being impossible, the combination of Rainforest and Katja Rembold for their support in the plant
sampling dominant species and the analysis with the identification. The authors thank Alexandra Dolynska and
fragment pool approach presents a method which is suit- Christine Radler for their help during lab work. The authors
able to calculate genetic differentiation of a plant express their appreciation to Tamar Weinberg and Ludger
community of different compositions with a very Leinemann for their suggestions on the manuscript. The
simple and universally applicable method. Even with a authors thank the Ministry of Research, Technology, and
low number of individuals the fragment pool approach is Higher Education for research permission in Indonesia. This
suitable to detect differences among plots and land-use study was conducted using samples based on the Collection
Permit No. 2697/IPH.1/KS.02/XI/2012 and 493/IPH.1/
systems regarding their genetic constitution.
KS.02.04/II/2015 recommended by the Indonesian Institute
Results of this study suggest that aside from forest,
of Science and issued by the Ministry of Forestry.
the investigated agroforest system jungle rubber also can
be an HCV. Jungle rubber cannot replace forests
(Rembold et al., 2017) but can present buffer zones Declaration of Conflicting Interest
and increase connectivity between HCV habitats and The author(s) declared no potential conflicts of interest with
facilitate gene flow (Orrock, 2005). The theoretical respect to the research, authorship, and/or publication of
design of Koh, Levang, and Ghazoul (2009) studies this article.
options for a compromise between conservational and
economic interests in a tropical landscape. The planned Funding
mosaic landscape contains forest fragments, continuous
The author(s) disclosed receipt of the following financial
forest, monocultures, and agroforestry parts. In the
support for the research, authorship, and/or publication of
overall landscape, biodiversity, ecosystem functions,
this article: The study was funded by the Deutsche
and human welfare are assured. Rules and indicators
Forschungsgemeinschaft (DFG), as part of EFForTS, CRC
for HCV habitats would give the possibility to design
landscapes with effective palm oil (or other crops) pro- (Collaborative Research Centre) 990 (https://www.uni-goettin
ductivity, buffer zones, and areas of natural forest. gen.de/de/310995). We acknowledge the support by the Open
Jungle rubber with high conservational value could be Access Publication Funds of the Georg-August Universit€at
included in such a designed landscape, but also agrofor- of G€ottingen.
ests ecosystems are vulnerable. Agroforest systems
depend on the economic advantages for the owner, ORCID iD
and, hence, older agroforests with higher conservational Natalie Breidenbach http://orcid.org/0000-0001-5339-4240
Breidenbach et al. 11
Goldman, D. H., Jansen, R. K., van den Berg, C., Leitch, I. J., Kashimshetty, Y., Pelikan, S., & Rogstad, S. H. (2015).
Fay, M. F., & Chase, M. W. (2004). Molecular and cyto- Variable gene dispersal conditions and spatial deforestation
logical examination of Calopogon (Orchidaceae, patterns can interact to affect tropical tree conservation
Epidendroideae): Circumscription, phylogeny, polyploidy outcomes. PloS One, 10, e0127745.
and possible hybrid speciation. American Journal of Kettle, C. (2014). Fragmentation genetics in tropical ecosys-
Botany, 91, 707–723. tems: From fragmentation genetics to fragmentation geno-
Goodman, D. (1975). The theory of diversity–stability relation- mics. Conservation Genetics, 15, 1265–1268.
ships in ecology. The Quarterly Review of Biology, 50, 237–266. Koh, L., Levang, P., & Ghazoul, J. (2009). Designer land-
Gouyon, A., Foresta, H., & Levang, P. (1993). Does “jungle scapes for sustainable biofuels. Trends in Ecology and
rubber” deserve its name? An analysis of rubber agroforest- Evolution, 24, 431–438.
ry systems in southeast Sumatra. Agroforestry Systems, 22, Koleff, P., Gaston, K., & Lennon, J. (2003). Measuring beta
181, 206. diversity for presence–absence data. Journal of Animal
Gower, J. (1966). Some distance properties of latent root and Ecology, 72, 367–382.
vector methods use in multivariate analysis. Biometrika, Kramer, A. T., Ison, J. L., Ashley, M. V., & Howe, H. F.
53, 325–338. (2008). The paradox of forest fragmentation genetics.
Grime, J. P. (2001). Plant strategies, vegetation processes, and Conservation Biology: The Journal of the Society for
ecosystem properties (2nd ed.). West Sussex, England: Conservation Biology, 22, 878–885.
John Wiley. Kramer, H., & Akca, A. (2008). Leitfaden zur Waldmesslehre
Hahn, T., Kettle, C. J., Ghazoul, J., Hennig, E. I., & Pluess, [Guide to forest gauging]. Frankfurt am Main, Germany: J.
A. R. (2013). Landscape composition has limited impact on D. Sauerl€ander Verlag.
local genetic structure in mountain clover, Trifolium mon- Kremer, A., Caron, H., Cavers, S., Colpaert, N., Gheysen, G.,
tanum L. The Journal of Heredity, 104, 842–852. Gribel, R., . . . Salgueiro, F. (2005). Monitoring genetic
Hamrick, J., Godt, M., & Sherman-Broyles, S. (1992). Factors diversity in tropical trees with multilocus dominant
markers. Heredity, 95, 274–280.
influencing levels of genetic diversity in woody plant species.
Kuchma, O. (2010). Genetic processes in Scots pine (Pinus syl-
New Forests, 6, 95–124.
vestris L.) in the Chernobyl exclusion zone (Unpublished
Hamrick, J. L., Linhart, Y. B., & Mitton, J. B. (1979).
doctoral dissertation). Georg-August-University,
Relationships between life history characteristics and elec-
G€ottingen, Germany.
trophoretically detectable genetic variation in plants.
Laliberté, E., & Tylianakis, J. (2012). Cascading effects of long-
Annual Review of Ecology and Systematics, 10, 173–200.
term land-use changes on plant traits and ecosystem func-
Hardwick, S. R., Toumi, R., Pfeifer, M., Turner, E. C., Nilus,
tioning. Ecology, 93, 145–155.
R., & Ewers, R. M. (2015). The relationship between
Lambin, E. F., Geist, H. J., & Lepers, E. (2003). Dynamics of
leaf area index and microclimate in tropical forest and oil
land-use and land-cover change in tropical regions. Annual
palm plantation: Forest disturbance drives changes in
Review of Environmental Resources, 28, 205–241.
microclimate. Agricultural and Forest Meteorology, Lande, R. (1988). Genetics and demography in biological con-
201, 187–195. servation. Science, 241, 1455–1460.
Hawthorne, W. D., & Abu-Juam, M. (1995). Forest protection Laumonier, Y. (1997). The vegetation and physiography of
in Ghana. With particular reference to vegetation and plant Sumatra. Dodrecht, the Netherlands: Kluwer
species. Gland, Schwitzerland: IUCN. Xvii þ 203pp. Academic Publishers.
Hendaryati, D. D., Adrianto, Y., Zuraina, W. K., Pudjianto, Laumonier, Y., Uryu, Y., Stüwe, M., Budiman, A., Setiabudi,
E., Udin, A., Kurniawati, N., . . . Magdalena, E. (2017). B., & Hadian, O. (2010). Eco-floristic sectors and defores-
Statistik Perkebunan Indonesia, Kelapa Sawit 2015–2017 tation threats in Sumatra: Identifying new conservation
[Statistics of Indonesian plantations, oil palm 2015–2017]. area network priorities for ecosystem-based land use plan-
Retrieved from http://ditjenbun.pertanian.go.id/tinymc ning. Biodiversity and Conservation, 19, 1153–1174.
puk/gambar/file/statistik/2017/Kelapa-Sawit-2015-2017.pdf Laurance, W. F., Sayer, J., & Cassman, K. G. (2014).
Honnay, O., & Jacquemyn, H. (2007). Susceptibility of common Agricultural expansion and its impacts on tropical nature.
and rare plant species to the genetic consequences of habitat Trends in Ecology and Evolution, 29, 107–116.
fragmentation. Conservation Biology, 21, 823–831. Laurance, W. F., Useche, D. C., Rendeiro, J., Kalka, M.,
Horn, H. S. (1966). Measurement of “overlap” in comparative Bradshaw, C. J. A., Sloan, S. P., . . . Zamzani, F. (2012).
ecological studies. American Naturalist, 100, 419–424. Averting biodiversity collapse in tropical forest protected
Hothorn, T., Bretz, F., & Westfall, P. (2008). Simultaneous areas. Nature, 489, 290–294.
inference in general parametric models. Biometrical Lee, S. H. J., Garcia-Ulloa, J., & Koh, L. P. (2011). Impacts of
Journal, 50, 346–363. biofuel expansion in biodiversity hotspots. In F. E. Zachos
Hughes, R., Inouye, B., Johnson, M., Underwood, N., & & J. H. Habel (Eds), Biodiversity hotspots. Distribution and
Vellend, M. (2008). Ecological consequences of genetic protection of conservation priority areas (pp.277–292).
diversity. Ecology Letters, 11, 609–623. Berlin, Germany: Springer.
Jacquemyn, H., De Meester, L., Jongejans, E., & Honnay, O. Leimu, R., Vergeer, P., Angeloni, F., & Ouborg, N. J. (2010).
(2012). Evolutionary changes in plant reproductive traits Habitat fragmentation, climate change, and inbreeding in
following habitat fragmentation and their consequences plants. Annals of the New York Academy of Science,
for population fitness. Journal of Ecology, 100, 76–87. 1195, 84–98.
Breidenbach et al. 13
Lesser, M. R., Parchman, T. L., & Jackson, S. T. (2013). Orrock, J. L. (2005). Conservation corridors affect the fixation
Development of genetic diversity, differentiation and struc- of novel alleles. Conservation Genetics, 6, 326–630.
ture over 500 years in four ponderosa pine populations. Parks, D. H., & Beiko, R. G. (2012). Measuring community
Molecular Ecology, 22, 2640–2652. similarity with phylogenetic networks. Molecular Biology
Lewontin, R. C. (1972). Testing the theory of natural selection. and Evolution, 29, 3947–3958.
Nature, 236, 181–182. Pinheiro, J., Bates, D., Debroy, D., Sarkar, D., & R Core
Li, J.-M., & Jin, Z.-X. (2007). Genetic structure of endangered Team. (2015). nlme: Linear and nonlinear mixed effect
Emmenopterys henryi Oliv. based on ISSR polymorphism models (R package Version 3.1-122) [Computer software].
and implications for its conservation. Genetica, 133, 227–234. Retrieved from http://cran.r-project.org/packages¼nlme
Linton, L. R., Davies, R. W., & Wrona, F. J. (1981). Resource Porter-Bolland, L., Ellis, E. A., Guariguata, M. R., Ruiz-
utilization indices: An assessment. The Journal of Animal Mallén, I., Negrete-Yankelevich, S., & Reyes-Garcıa, V.
Ecology, 50, 283–292. (2012). Community managed forests and forest protected
Lynam, T., de Jong, W., Sheil, D., Kusumanto, T., & Evans, K. areas: An assessment of their conservation effectiveness
(2007). A review of tools incorporating community knowl- across the tropics. Forest Ecology and Management,
edge, preferences, and values into decision making in natural 268, 6–17.
resources management. Ecology and Society, 12, 5. Rametsteiner, E., & Simula, M. (2003). Forest certification—
Mace, E., Gebhardt, C., & Lester, R. (1999). AFLP analysis of An instrument to promote sustainable forest management?
genetic relationships in the tribe Datureae (Solanaceae). Journal of Environmental Management, 67, 87–98.
Theoretical and Applied Genetics, 99, 634–641. Reed, B. M., Sarasan, V., Kane, M., Bunn, E., & Pence, V. C.
Margono, B. A., Potapov, P. V., Turubanova, S., Stolle, F., & (2011). Biodiversity and conservation biotechnology tools.
Hansen, M. C. (2014). Primary forest cover loss in Indonesia In Vitro Cellular and Developmental Biology-Plant, 47, 1–4.
over 2000–2012. Nature Climate Change, 4, 730–735. Rembold, K., Mangopo, H., Tjitrosoedirdjo, S., & Kreft, H.
Mba, C., & Tohme, J. (2005). Use of AFLP markers in surveys (2017). Plant diversity, forest dependency, and alien plant
of plant diversity. Methods in Enzymology, 395, 177–201.
invasions in tropical agricultural landscapes. Biological
Meudt, H. M., & Clarke, A. C. (2007). Almost forgotten or
Conservation, 213, 234–242.
latest practice? AFLP applications, analyses and advances.
Robinson, J. P., & Harris, S. A. (1999). Amplified fragment
Trends in Plant Science, 12, 106–117.
length polymorphisms and microsatellites: A phylogenetic
Michon, G., & Foresta, D. H. (1995). Agroforests: An original
perspective. In E. Gillet (Ed.), Which DNA marker for
agroforestry model from smallholder farmers for environ-
which purpose? Molecular tools for biodiversity.
mental conservation and sustainable development. In K.
Hamburg, Germany: Bundesforschunganstalt für Forst-
Ishizuka, S. Hisajima, & D. R. J. Macer (Eds), Traditional
und Holzwirtschaft, Hamburg, Germany. ISSN: 0368-8798
technology for environmental conservation and sustainable
Rodrıguez-Echeverry, J., Echeverrıa, C., Oyarz un, C., &
development in the Asian-Pacific Region. Tsukuba, Japan,
Morales, J. (2018). Impact of land-use change on biodiver-
University of Tsukuba.
sity and ecosystem services in the Chilean temperate forests.
Mona, S., Ray, N., Arenas, M., & Excoffier, L. (2014). Genetic
consequences of habitat fragmentation during range expan- Landscape Ecology, 33, 439–453.
Roundtable on Sustainable Palm Oil. (2013, April). Principles
sion. Heredity, 112, 291–299.
Myers, N. (1988). Threatened biotas: “Hot spots” in tropical and criteria for the production of sustainable palm oil.
forests. Environmentalist, 8, 187–208. Endorsed by the RSPO Executive Board and Accepted at
Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, the Extraordinary General Assembly by RSPO members.
G. A. B., & Kent, J. (2000). Biodiversity hotspots for con- Sampson, J. F., Byrne, M., Yates, C. J., Gibson, N.,
servation priorities. Nature, 403, 853–858. Thavornkanlapachai, R., Stankowski, S., . . . Bennett, I.
Newbold, T., Hudson, L. N., Hill, L. L., Contu, S., Lysenko, (2014). Contemporary pollen-mediated gene immigration
I., Senior, R. A., . . . Purvi, A. (2015). Global effects of land reflects the historical isolation of a rare, animal-pollinated
use on local terrestrial biodiversity. Nature, 520, 45–49. shrub in a fragmented landscape. Heredity, 112, 172–181.
Noordwijk, M., Suyamto, D., Lusiana, B., Ekadinata, A., & Savilaakso, S., Garcia, C., Garcia-Ulloa, J., Ghazoul, J.,
Hairiah, K. (2008). Facilitating agroforestation of land- Groom, M., Guariguata, M. R., .. . Zrust, M. (2014).
scapes for sustainable benefits: Tradeoffs between carbon Systematic review of effects on biodiversity from oil palm
stocks and local development benefits in Indonesia accord- production. Environmental Evidence, 3(1), 20.
ing to the FALLOW model. Agriculture, Ecosystems and Schroth, G., Harvey, C. A., & Vincent, G. (2004). Complex
Environment, 126, 98–112. Agroforests: Their structure, diversity, and potential role in
Oksanen, J., Blanchet, F. G., & Kindt, R. (2015). Package landscape conservation. In G. Schroth, G. A. B. da
“vegan.” Community ecology package (Rpackage Version Fonseca, C. A. Harvey, C. Gascon, H. L. Vasconcelos, &
2.3-0) [Computer software]. Retrieved from http://cran.r- A-M. N. Izac (Eds), Agroforestry and biodiversity conserva-
project.org/package¼vegan tion in tropical landscapes. Washington, DC: Island Press.
Ootsuki, R., Sato, H., Nakato, N., & Murakami, N. (2012). Sebbenn, A., Degen, B., Azevedo, V., Silva, M., de Lacerda,
Evidence of genetic segregation in the apogamous fern spe- A., Ciampi, A., . . . Loveless, M. (2008). Modelling the long-
cies Cyrtomium fortunei (Dryopteridaceae). Journal of Plant term impacts of selective logging on genetic diversity
Research, 125, 605–612. and demographic structure of four tropical tree species in
14 Tropical Conservation Science
the Amazon forest. Forest Ecology and Management, Jambi (Sumatra), Indonesia. Regional Environmental
254, 335–349. Change, 14, 825–834.
Shannon, C. E., & Weaver, W. (1949). The mathematical theory Violle, C., Enquist, B., McGill, B., Jiang, L., Albert, C.,
of communication. Urbana: University of Illinois Press. Hulshof, C., . . . Messier, J. (2012). The return of
Sloan, S., Jenkins, C. N., Joppa, L. N., Gaveau, D., & variance: Intraspecific variability in community ecology.
Laurance, W. F. (2014). Remaining natural vegetation in Trends in Ecology and Evolution, 27, 244–252.
the global biodiversity hotspots. Biological Conservation, Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T.,
177, 12–24. Homes, M., . . . Zabeau, M. (1995). AFLP: A new technique
Sodhi, N. S., Koh, L. P., Brook, B., & Ng, P. (2004). Southeast for DNA fingerprinting. Nucleic Acids Research,
Asian biodiversity: An impending disaster. Trends in 23, 4407–4414.
Ecology and Evolution, 19, 654–660. Vranckx, G., Jacquemyn, H., Muys, B., & Honnay, O. (2011).
Sodhi, N. S., Posa, M. R. C., Lee, T. M., Bickford, D., Koh, Meta-analysis of susceptibility of woody plants to loss of
L. P., & Brook, B. W. (2010). The state and conservation of genetic diversity through habitat fragmentation.
Southeast Asian biodiversity. Biodiversity and Conservation, Conservation Biology, 26, 228–237.
19, 317–328. Wang, I. J., & Bradburd, G. S. (2014). Isolation by environ-
Souto, C. P., Mathiasen, P., Acosta, M. C. C., Quiroga, ment. Molecular Ecology, 23, 5649–5662.
M. P. P., Vidal-Russell, R., Echeverrıa, C., & Premoli, White, M. J. (1986). Segregation and diversity measures in
A. C. (2015). Identifying genetic hotspots by mapping population distribution. Population Index, 52, 198–221.
molecular diversity of widespread trees: When commonness Whitham, T., Young, W. P., Martinsen, G. D., Gehring, C. A.,
Schweitzer, J. A., Shuster, S. M., . . . Kus, C. R. (2003).
matters. Journal of Heredity, 537–545.
Community and ecosystem genetics: A consequence of the
Stephens, M. (1979). Tests of fit for the logistic distribution
extended phenotype. Ecology, 84, 559–573.
based on the empirical distribution function. Biometrika,
Whitham, T., Bailey, J. K., Schweitzer, J. A., Shuster, S.,
66, 591–595.
Bangert, R. K., LeRoy, C. J., . . . Wooley, S. C. (2006).
Sthultz, C., Gehring, C., & Whitham, T. (2009). Deadly com-
A framework for community and ecosystem genetics:
bination of genes and drought: Increased mortality of
From genes to ecosystems. Nature Reviews Genetics,
herbivore-resistant trees in a foundation species. Global
7, 510–523.
Change Biology, 15, 1949–1961. Wickham, H. (2007). Reshaping data with the reshape pack-
Stoll-Kleemann, S. (2010). Evaluation of management effec- age. Journal of Statistical Software, 12, 1–20.
tiveness in protected areas: Methodologies and results. Wickham, H. (2009). ggplot2: Elegant graphics for data analy-
Basic and Applied Ecology, 11, 377–382. sis. New York, NY: Springer.
Symes, W. S., Rao, M., Mascia, M. B., & Carrasco, R. L. Wilcove, D., Giam, X., Edwards, D., Fisher, B., & Koh, L. P.
(2015). Why do we lose protected areas? Factors influencing (2013). Navjot’s nightmare revisited: Logging, agriculture,
protected area downgrading, downsizing and degazettment and biodiversity in Southeast Asia. Trends in Ecology and
(PADDD) in the tropics and sub-tropics. Global Change Evolution, 28, 531–540.
Biology, 22, 656–665. Wilcove, D., & Koh, L. P. (2010). Addressing the threats to
Thomas, E., Jalonen, R., Loo, J., Boshier, D., Gallo, L., biodiversity from oil-palm agriculture. Biodiversity and
Cavers, S., . . . Bozzano, M. (2014). Genetic considerations Conservation, 19, 999–1007.
in ecosystem restoration using native tree species. Forest Wolda, H. (1981). Similarity indices, sample size and diversity.
Ecology and Management, 333, 66–75. Oecologia, 50, 296–302.
Thompson, I. D., Guariguata, M. R., Okabe, K., Yeh, F. C., & Boyle, T. J. B. (1997). Population genetic anal-
Bahamondez, C., Nasi, R., Heymell, V., & Sabogal, C. ysis of co-dominant and dominant markers and quantitative
(2013). An operational framework for defining and moni- traits. Belgian Journal of Botany, 129, 157.
toring forest degradation. Ecological Society, 18, 20. Young, A., Boyle, T., & Brown, T. (1996). The population
Villamor, G., Pontius, R., Jr., & Noordwijk, M. (2013). genetic consequences of habitat fragmentation for plants.
Agroforest’s growing role in reducing carbon losses from Trends in Ecology and Evolution, 11, 413–418.