Global Ecology and Conservation 10 (2017) 173–183

Contents lists available at ScienceDirect

Global Ecology and Conservation

journal homepage: www.elsevier.com/locate/gecco

Original research article

Functional diversity in a fragmented landscape — Habitat

alterations affect functional trait composition of frog
assemblages in Madagascar
Jana C. Riemann a, *, Serge H. Ndriantsoa b , Mark-Oliver Rödel c , Julian Glos a
a
Zoological Institute, Animal Ecology and Conservation, University of Hamburg, Martin-Luther-King Platz 3, 20146 Hamburg, Germany
b
Mention Zoologie et Biodiversité Animale, Faculté des Sciences, Université d’Antananarivo, Antananarivo 101, Madagascar
c
Museum für Naturkunde, Leibniz Institute for Evolution and Biodiversity Science, Invalidenstr. 43, 10115 Berlin, Germany

highlights

• Habitat fragmentation and conversion altered resource use trait composition.

• The relationship between functional and species richness differed between habitats.
• Altered habitats may not always provide less, but different functions.
• Functional clustering occured in altered and natural habitats.
• Functional aspects need to be considered for conservation planning.

article info a b s t r a c t
Article history: Anthropogenic habitat alterations cause biodiversity loss, which in turn negatively affects
Received 1 November 2016 ecosystem functioning and services, and thus human well-being. To be able to consider
Available online 11 April 2017 ecosystem functioning in conservation actions, analyzing the effects of habitat alteration
on functional diversity is essential. Some altered habitats can maintain a significant part
Keywords:
of regional biodiversity, however, functional diversity information in altered habitats
Functional clustering
is so far mostly lacking. We compared functional richness and functional β -diversity
Functional ß-diversity
Ecosystem processes based on resource-use traits of frogs between three land-use categories in a rainforest
Ecosystem functioning ecosystem in Madagascar. Land-use categories represent a habitat alteration gradient
Rainforest fragmentation ranging from continuous forest over forest fragments to matrix habitats including different
Amphibian conservation agricultures. Our study revealed distinct changes in resource-use trait composition and
complex patterns in the relationship between species richness and functional richness.
Thus, the functional structure of frog assemblages changed due to habitat alterations.
However, altered habitats likely provide different, rather than fewer functions compared
to intact forest. Streams in all land-use categories were the functionally richest habitats,
and thus important for ecosystem functioning. Species richness was one, but not the
only driver of functional richness in our system. Functional clustering, potentially due to
environmental filters depending on resource availability, was caused by anthropogenic
and natural drivers. Our study shows that, even in systems where fragmented landscapes
still maintain high species diversity, functional diversity can be altered in human altered

*
Corresponding author.
E-mail addresses: jana.riemann@uni-hamburg.de, jcriemann@gmail.com (J. Riemann), nsehel2006@gmail.com (S. Ndriantsoa),
mo.roedel@mfn-berlin.de (M.-O. Rödel), julian.glos@uni-hamburg.de (J. Glos).

http://dx.doi.org/10.1016/j.gecco.2017.03.005
2351-9894/© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
licenses/by-nc-nd/4.0/).
174 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183

habitats, which may affect ecosystem processes like productivity, nutrient cycling, and
energy flows.
© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Anthropogenic habitat alterations like deforestation, fragmentation, and conversion into agricultural land are threatening
biodiversity globally (Butchart et al., 2010; Dirzo and Raven, 2003). Biodiversity loss in turn can affect ecosystem processes
and functioning and, as a consequence, ecosystem services with negative consequences for humans (Cardinale et al., 2012). It
is thus of concern to reveal processes shaping communities under disturbances and link consequences of habitat alterations
and biodiversity loss to ecosystem functioning, which can be achieved using trait based functional approaches (Cadotte et
al., 2011; Mouillot et al., 2013). Functional diversity, measured as the extent of functional trait differences among species in
a community, is the component of biodiversity that influences ecosystem processes (e.g., productivity, resource dynamics)
relevant for ecosystem functioning (i.e., how an ecosystem operates) (Tilman, 2001). Functional traits are traits that influence
organismal performance (Mouillot et al., 2013), and need to reflect the ecosystem processes of interest (Petchey and Gaston,
2006, 2002). Understanding the effects of anthropogenic habitat alterations on functional diversity is essential to direct
conservation efforts beyond pure species numbers, towards a more inclusive approach prioritizing ecosystem function and
stability (Cadotte et al., 2011).
Anthropogenic habitat alterations can affect functional diversity through different mechanisms. One important driver of
loss in functional diversity is species loss, however deviations from this relationship are likely most relevant for effects on
ecosystem functioning (Flynn et al., 2009). If species richness declines in altered habitats, functional diversity may remain
constant if assemblages show high redundancy. Alternatively, functional diversity may decline with species richness, but in
different ways. A reduction in functional diversity may either simply reflect habitat alteration effects on species richness, or
assemblages that are functionally more similar (clustered or redundant) than expected in relation to the number of species
if environmental trait filtering operates. These processes need to be separated to fully understand the mechanisms driving
potential reductions in functional diversity in altered habitats (Cadotte et al., 2011; Flynn et al., 2009; Mouillot et al., 2013).
Given current deforestation rates (Hansen et al., 2013) and agricultural expansions (Laurance et al., 2014), conservation of
tropical biodiversity will become more and more dependent on appropriately managed human-modified landscapes. There
is increasing evidence that forest fragments and, depending on habitat quality or availability of structural elements, even
matrix or agricultural habitats can still maintain a significant part of regional taxonomic diversity and may thus contribute
to biodiversity conservation (Gillespie et al., 2012; Mendenhall et al., 2014; Ndriantsoa et al., 2017; Nopper et al., 2017;
Riemann et al., 2015). However, it is unclear whether these habitats can also maintain functional diversity comparable
to natural habitats. Human habitat modifications and land-use intensification reduce functional diversity of various taxa
(Almeida et al., 2016; Flynn et al., 2009; Mumme et al., 2015).
So far only few studies addressed functional diversity of amphibians in human altered habitats (e.g., Ernst et al.,
2006). Globally, many amphibian species are threatened by multiple, often interacting stressors, but especially habitat
loss (Chanson et al., 2008). Larval as well as juvenile and adult anurans are important prey for various taxa including
invertebrates, fishes, amphibians, reptiles, birds and mammals (Toledo, 2005; Toledo et al., 2007; Wells, 2007), but are also
effective predators of many taxa (Altig et al., 2007; Duellman and Trueb, 1994). Given their relevance and contributions
to various ecosystem processes, vanishing amphibians are likely to trigger cascading effects on ecosystem functions and
services (Davic and Welsh, 2004; Hocking and Babbitt, 2014; Whiles et al., 2006). It is thus vital to evaluate and understand
potential ecosystem consequences of the current global amphibian decline (Cox et al., 2008).
We focused on the effects of forest fragmentation and conversion on frog functional diversity in Madagascar to reveal
possible habitat alteration effects on amphibian assemblages beyond effects on taxonomic diversity. Functional diversity
analyses can reveal habitat alteration effects on biodiversity beyond effects on species richness and composition that often
remain undetected by classical diversity measures and further allow for drawing inferences about potential consequences
for ecosystem processes and functions (Cadotte et al., 2011; Mouillot et al., 2013). Madagascar belongs to the world’s
biodiversity hotspots with an outstanding degree of endemism (Goodman and Benstead, 2005; Myers et al., 2000), especially
for frogs (Glaw and Vences, 2007). Madagascar’s natural ecosystems are highly threatened by deforestation, forest
fragmentation, and human land-use intensification (Allnutt et al., 2008; Harper et al., 2007). Understanding the effects of
habitat alterations on biodiversity is essential for long-term success of conservation efforts (Irwin et al., 2010), especially
since human populations highly depend on the resources and ecosystem services natural ecosystems provide (Kari and
Korhonen-Kurki, 2013).
We investigated two aspects of functional diversity using a set of twelve ecological, morphological and life history
traits related to resource use. Resource-use traits are expected to reflect the relationship between functional diversity and
ecosystem functioning (Cadotte et al., 2011; Petchey and Gaston, 2006). Our selected traits are proxies for where, when,
which, and in what quantity frogs exploit resources for growth, reproduction, and survival, but in turn also determine
where and by whom frogs can be used as food resource. We consider these traits to be associated with ecosystem processes
like productivity, nutrient cycling, and energy flows, the major ecosystem functions amphibians are involved in (Hocking
 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183 175

and Babbitt, 2014). We compared functional richness and functional β -diversity of frogs between three land-use categories
that represent an anthropogenic habitat alteration gradient ranging from continuous forest over forest fragments to matrix
habitat including agriculture (banana plantations and rice fields) in a mid-altitude rainforest ecosystem. We differentiated
between changes in functional richness that were directly linked to changes in species richness and changes in functional
richness independent from species richness. We hypothesized that (1) functional richness is reduced in forest fragments
and especially in matrix habitats compared to continuous forest, and that forest fragmentation and conversion (2) affect
functional richness via effects on species richness and beyond, and (3) cause distinct patterns of functional β -diversity, i.e., a
turnover in trait composition, between continuous forest, forest fragments, and matrix habitats.

2. Material and methods

2.1. Study system

Field work was conducted in the Ranomafana region, southeast Madagascar, a hotspot of amphibian diversity that harbors
about 120 species (Glaw and Vences, 2007; Vieites et al., 2009; own unpubl. data). Ranomafana National Park (RNP, 21◦ 02’–
21◦ 24’S, 47◦ 20’–47◦ 35’E, 43 500 ha) contains most of the remaining mid-altitude rainforest (500–1300 m a.s.l.) habitat in
that area and is surrounded by a highly fragmented landscape. Few remaining forest fragments are embedded in a matrix
of cultivated land (slash and burn agriculture; ‘‘tavy’’), and secondary bush and shrub vegetation on fallow land. The major
crops grown are rice and bananas, and to a lesser extent cassava, coffee, and other fruits and vegetables. The nine studied
forest fragments ranged in size between two and 16.5 ha. A more comprehensive description of the study site is provided
in Riemann et al. (2015) and Ndriantsoa et al. (2017).

2.2. Species assemblage data

We used species occurrence data from Riemann et al. (2015) and Ndriantsoa et al. (2017) that were collected on 56 line
transects (50 × 2 m; Marsh and Haywood, 2010) spread over three land-use categories that represent the anthropogenic
disturbance gradient continuous forest (CF; control sites inside RNP) – forest fragments (FF) – matrix (MX). We considered
stream (s) and terrestrial (t) habitats in all three land-use categories to cover stream-dependent species (either stream
breeding or semiaquatic) as well as species that reproduce independent from streams (i.e., phytotelmata, pond or terrestrial
breeders). Additionally, banana plantations (b) and rice fields (r) were included as these habitats represent the most
prominent agriculture in the matrix. Sample sizes varied between these eight habitat types according to availability (missing
permission from farmers to access more matrix sites): CFs (11), CFt (11), FFs (6), FFt (10), MXs (5), MXt (4), MXb (4), MXr (5).
Stream transects included one meter riparian vegetation on each stream bank and the stream. We kept a minimum
distance of 50 m between terrestrial and stream transects and 200 m between transects of the same habitat type. Terrestrial
transects in the matrix were characterized by secondary vegetation (e.g. ferns, grasses, shrubs, guava trees) on fallow land.
Stream transects in the matrix were lined with approximately one to five meter narrow riparian vegetation on each stream
bank, consisting of trees and bushes. Rice fields always contained or were directly adjacent to streams, and water was
constantly provided through small irrigation ditches. Most of their area was constantly inundated during the rainy season.
Banana plantations usually contained an understory herb layer and few forest trees, as well as small non-permanent rills or
irrigation ditches.
We conducted standardized visual and acoustic transect sampling to detect all major ecological frog guilds (Rödel and
Ernst, 2004; Vences et al., 2008). However, we may have missed species with weak calls inhabiting tree holes, Pandanus
or Ravenala trees, and did not particularly target fossorial species. All transects were visited nine times (once by day and
once by night from May to June in 2010; three times by day and four times by night between January and June in 2011)
alternately by two teams. Habitat types were sampled alternately to avoid temporal effects on sampling results. Transects
were sampled at constant speed (approx. 2.5 m/min) covering all visible individuals from leaf litter up to arboreal structures
in approximately 2.5 m height. Visually detected individuals were captured, measured, sexed and marked via toe clipping
(no functionally important toes, following the recommendations by Grafe et al., 2011; toe clips were collected as tissue
samples for further analyses). Except of voucher specimens needed for accurate species identification (deposited at Mention
Zoologie et Biodiversité Animale, Université d’Antananarivo, Madagascar, and the Museum für Naturkunde Berlin, Germany;
see Riemann et al., 2015; Ndriantsoa et al., 2017), all individuals were immediately released at the point of capture. Calling
individuals were recorded for five minutes at each of four fixed points (start, 12.5, 25, 37.5 m) in a 12.5 m distance (Rödel
and Ernst, 2004) straight, left and right to the transect. To ascertain the completeness of local species assemblages, we
additionally conducted opportunistic surveys at each site. Species were identified by morphology (according to Glaw and
Vences, 2007) and advertisement calls (according to Vences et al., 2006; own reference data) in the field and confirmed by
DNA barcoding based on a fragment of the mitochondrial 16S rRNA gene (Vences et al., 2008, 2005). Obtained sequences
were compared to published data on GenBank or own reference sequences (see Riemann et al., 2015; Ndriantsoa et al., 2017
for GenBank accession numbers).
Species recorded only once during the study period (Gephyromantis depressiceps, Plethodontohyla notosticta) were
excluded from the data set. The resulting species presence/absence matrix contained 62 species which were considered
as species pool for analyses. A more detailed description of the sampling design is provided in Riemann et al. (2015) and
Ndriantsoa et al. (2017).
176 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183

Table 1
Selected traits associated with resource use and the chosen characteristic attributes for adult anurans. The decisive webbing formula (after Glaw and
Vences, 1994) for trait foot webbing is given in parentheses. Reproductive modes are classified according to Duellman and Trueb (1994) (missing numbers
refer to reproductive modes that are not present in the sampled frog assemblage) and were assigned to species following Glaw and Vences (2007).
Reproductive mode 15 here includes nidicolous tadpoles according to Randrianiaina et al. (2011).
Trait Scale Characteristic attributes / Unit
Daily activity Nominal 1: diurnal, 2: nocturnal, 3: cathemeral
Basic habitat Nominal 1: semiaquatic, 2: riparian ground incl. vegetation (veg) ≤ 1 m, 3:
terrestrial incl. veg ≤ 1 m, 4: arboreal incl. tree holes, 5: semiarboreal
Calling position Nominal 1: semiaquatic, 2: riparian ground incl. veg ≤ 1 m, 3: terrestrial incl.
veg ≤ 1 m, 4: arboreal incl. tree holes, 5: semiarboreal
Mean male snout vent length Numeric Mean [mm]
Mean male head width Numeric Mean [mm]
Relative hind limb length Ordinal 1: < tymp, 2: tymp –eye, 3: eye –snout, 4: beyond snout
Hand webbing Ordinal 1: absent, 2: trace or not between all fingers, 3: present between all
fingers
Foot webbing Ordinal 1: absent or traces (4i/e > 3), 2: medium (4i/e > 1 ≤ 3), 3: full (4i/e ≤
1)
Terminal disks Ordinal 1: not expanded, 2: slightly expanded, 3: expanded
Sex size dimorphism Nominal 0: no, 1: yes
Reproductive mode Nominal 1: eggs and feeding tadpoles in lentic water, 2: eggs and feeding
tadpoles in lotic water, 6: eggs and non-feeding tadpoles in water in
tree holes or aerial plants, 12: eggs and early tadpoles in excavated
nest (on ground); subsequent to flooding, feeding tadpoles live in
ponds or streams, 15: eggs hatch into non-feeding tadpoles that
complete their development in nest (on ground), 18: eggs (arboreal)
hatch into tadpoles that drop into ponds or streams, 22: nest in
burrow; non-feeding tadpoles complete development in foam nest
Basic coloration Nominal 1: green, 2: brown, 3: green/brown, 4: multicolored

2.3. Trait data

We selected twelve ecological, morphological and life history traits that are related to resource-use including nominal,
ordinal, and numeric variables (Table 1). Trait data for all 62 species from the considered species pool (Appendix A1) was
gained from different sources: (i) collected during transect walks and opportunistic surveys, (ii) measured on collected
voucher specimens (see 2.2), and (iii) obtained from field guides by Glaw and Vences (2007, 1994) and additional literature
(Glaw et al., 2010; Glaw and Thiesmeier, 1993; Vallan et al., 2010; Vences and Glaw, 2004, 2002; Vences et al., 2012).

2.4. Data analyses

We used trait matrix A1 (Appendix A1, Supplementary data) to analyze the effects of forest fragmentation and conversion
on functional richness and functional β -diversity based on species presence/absence data. All statistical analyses were
performed in R version 3.2.2 (R Core Team, 2015).
We used the FD measure by Petchey and Gaston (2006, 2002) as a measure of functional richness (Mouchet et al., 2010).
FD is defined as the total branch length of a dendrogram excluding the root segment and is calculated in three steps. First,
the trait matrix was converted into a distance matrix (pairwise distances between species in functional trait space) based
on Gower‘s distance to handle variables of mixed type with function ‘‘daisy’’ in R package cluster (Maechler et al., 2015).
A dendrogram (Appendix A2) was then built by hierarchical clustering using the group average method (UPGMA) which
produced the highest correlation between the original distances and the cophenetic distances (Mantel statistic based on
Pearson’s product-moment correlation, r = 0.826, p = 0.001, number of permutations: 999). The high correlation indicates
that the data were appropriately represented by the dendrogram. We finally used function ‘‘treedive’’ in R package vegan
(Oksanen et al., 2016) to calculate observed FD, i.e. observed functional richness (FDobs ) per transect, which is the total branch
length of the dendrogram subsets for each species assemblage. We used Kruskal–Wallis rank sum analyses of variance to
compare FDobs between habitats. P-values of post-hoc pairwise comparisons using Wilcoxon rank sum test were adjusted
according to Benjamini and Hochberg (1995). Spearman’s rank correlation was used to assess the correlation between
species richness and FDobs in the whole data set.
Since the FD measure is generally positively correlated with species richness (Mouchet et al., 2010), we used a null
model approach to control for differences in species richness and evaluate if observed functional richness deviates from
expected values to reveal processes beyond species loss that shape functional diversity in altered habitats. We randomized
functional trait data by shuffling of species names on the tips of the dendrogram, analog randomization of phylogenetic
data according to Swenson (2014). This approach preserves species richness and all assemblage patterns in the species
assemblage matrix but randomizes similarity of species in trait space and hence species’ functional diversity. We calculated
a FD null distribution for each transect with 999 randomizations (Appendix A3). We used Wilcoxon signed rank test to
evaluate deviations of FDobs from randomly expected functional richness (mean of null distribution, FDexp ) per habitat type.
 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183 177

Fig. 1. Observed functional richness in different habitats (stream: s, terrestrial: t, banana plantation: b, rice field: r) in the land-use gradient continuous
forest (CF, dark gray boxes) – forest fragments (FF, light gray boxes) – matrix (MX, white boxes). Shown are median, inter-quartile range (box), and location
of the minimum and maximum (whiskers). Different italic small letters indicate significant differences (p < 0.05) between habitats according to post-hoc
pairwise comparisons using Wilcoxon rank sum test.

To assess direction and magnitude of deviations from null model expectations we calculated standardized effect sizes (SES)
as SES = (FDobs − FDexp ) / sd null distribution (Swenson, 2014). Positive SES represent a higher FDobs than FDexp (functional
overdispersion), negative SES represent a lower FDobs than FDexp (functional clustering). We used Kruskal Wallis rank sum
analyses of variance to compare SES between habitats. P-values of post-hoc pairwise comparisons using Wilcoxon rank sum
test were adjusted according to Benjamini and Hochberg (1995).
We chose a dendrogram or tree based measure to calculate functional β -diversity (β -FD) to be coherent in methodology
and allow for inferences from functional richness (FD) and β -FD. We used the unweighted UniFrac metric of Lozupone
and Knight (2005), a phylogenetic β -diversity metric which calculates the unique fraction (non-shared branches) of total
phylogenetic diversity (branch-lengths) between two assemblages. It can be used to calculate functional β -diversity if
the dendrogram or tree is derived from functional trait data (Swenson, 2014). UniFrac is a distance metric that fulfills
all requirements to be used with standard multivariate statistics (Lozupone and Knight, 2005). We calculated pairwise
UniFrac with function ‘‘unifrac’’ in R package picante (Kembel et al., 2010) for all transect combinations. This required the
transformation of the dendrogram in a tree of class ‘‘phylo’’ which did not change branch lengths (Mantel statistic based on
Pearson’s product-moment correlation, r = 1, p = 0.001, number of permutations: 999) or any pattern in the dendrogram.
We used non-metric multidimensional scaling (NMDS) to visualize and evaluate patterns of dissimilarity in trait
composition among transects based on the UniFrac metric (β -FD). NMDS was performed with function ‘‘metaMDS’’ from
R package vegan (Oksanen et al., 2016). This method uses random starting configurations to find a stable global solution.
We used function ‘‘adonis’’ from R package vegan (Oksanen et al., 2016) to perform permutational multivariate analyses
of variance (perMANOVA) (Anderson, 2001) in order to test whether β -FD differs within the habitat alteration gradient
continuous forest – forest fragments – matrix. This non-parametric permutation based variant of MANOVA partitions sums
of squares of multivariate data equivalent to univariate ANOVA and the pseudo F statistic can be calculated directly from
any distance measure (Anderson, 2001). The p-value was obtained from 999 permutations.

3. Results

3.1. Functional richness

Observed functional richness (FDobs ) was significantly higher in all stream habitats compared to terrestrial and agricul-
tural habitats, and was lowest in terrestrial matrix habitat that differed significantly from all other habitat types (Kruskal-
Wallis ANOVA, X2 7 = 44.14, p < 0.001, Fig. 1). However, FDobs neither differed between stream habitats in different land-use
categories, nor between terrestrial habitats in continuous forest and forest fragments, and between continuous forest and
fragment terrestrial habitats and the two cultivated habitats (i.e. banana plantation and rice fields, Fig. 1). We observed a
strong positive correlation between FDobs and species richness in the whole data set (rS = 0.94, p < 0.001), indicating that
differences in species richness highly influenced FDobs (Fig. 2).
Standardized effect sizes (SES) revealed low, but distinct negative deviations in FDobs from null model expectations (FDexp )
in terrestrial habitat in continuous forest and all matrix habitats except rice fields (Fig. 3). In these habitats assemblages were
178 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183

Fig. 2. Correlation between observed functional richness and species richness considering all habitat types (stream: circle, terrestrial: triangle, banana
plantation: diamond, rice field: square) in the land-use gradient continuous forest (dark gray) – forest fragments (light gray) – matrix (white).

functionally more clustered than randomly expected in relation to the observed species richness, indicating that species
richness is not the only driver of functional richness in our system. However, FDobs significantly differed from FDexp only
in terrestrial habitat in continuous forest (Wilcoxon signed rank test, V = 0, p = 0.001, n = 11). The observed trends in
the matrix were not supported statistically, potentially because small sample sizes for matrix habitats reduced statistical
power. Deviations from null model expectations, i.e. SES, differed between habitats (Kruskal-Wallis ANOVA, X2 7 = 18.015,
p = 0.012, Fig. 3). Post-hoc pairwise comparisons using Wilcoxon rank sum test revealed a significant difference of terrestrial
habitat in continuous forest from streams in continuous forest (p = 0.029) and forest fragments (p = 0.043), all other
pairwise comparisons did not reveal any significant differences.

3.2. Functional β -diversity

We found distinct effects of forest fragmentation and conversion on functional β -diversity (β -FD) as trait composition
significantly differed within the land-use gradient continuous forest – forest fragments – matrix (perMANOVA, F7,48 = 7.55,
R2 = 0.52, p = 0.001). NMDS revealed distinct patterns of dissimilarities in β -FD between transects according to their
habitat type (Fig. 4). The stress value (0.16) of the final two-dimensional solution indicated a reasonable preservation
of ordering relationships of the multidimensional among-transect dissimilarities. Habitats were separated along axis
one according to water availability, i.e. the presence of streams or ditches. The separation along axis two reflected the
anthropogenic habitat alteration gradient continuous forest – forest fragments – matrix, with rice fields being most separated
from all other habitats.

4. Discussion

Madagascar’s biodiversity is threatened by ongoing habitat loss and fragmentation (Harper et al., 2007). Although in
general biodiversity responds negatively to these anthropogenic disturbances, reactions vary greatly between taxa and
ecoregions (Irwin et al., 2010). It has been shown that rainforest fragments and even banana plantations and streams in
the matrix can still be suitable habitats for many frog species and may contribute to their conservation (Ndriantsoa et al.,
2017; Riemann et al., 2015). However, so far information about the persistence of functional diversity in human altered
landscapes is lacking. Understanding fragmentation effects on functional diversity is crucial for future conservation planning
as it allows a more inclusive perspective of habitat alteration effects on biodiversity than measures based on species numbers
and composition alone and further allows inferences about potential effects on ecosystem processes (Cadotte et al., 2011;
Mouillot et al., 2013).
Our study revealed changes in the functional structure of frog assemblages using resource-use traits, i.e. complex patterns
of functional richness and differences in functional ß-diversity, in the land-use gradient ranging from continuous forest
over forest fragments to the matrix. Although functional richness usually declines in altered habitats (e.g., Almeida et al.,
2016; Ernst et al., 2006), it was significantly reduced in only one human altered habitat type (terrestrial matrix habitat)
compared to respective habitats in continuous forest in our study. Thus, habitat alterations do not always affect functional
richness. However, identical functional richness values can be caused by different trait combinations. As expected, patterns
of functional β -diversity reflected the anthropogenic disturbance gradient and additionally water availability. We detected
 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183 179

Fig. 3. Standardized effect sizes (SES) showing deviations in observed functional richness (FDobs ) from null model expectations (FDexp ) for different habitats
(stream: s, terrestrial: t, banana plantation: b, rice field: r) in the land-use gradient continuous forest (CF, dark gray boxes) – forest fragments (FF, light gray
boxes) – matrix (MX, white boxes). Positive SES represent a higher FDobs than FDexp , negative SES represent a lower FDobs than FDexp . Shown are median,
inter-quartile range (box), and location of the minimum and maximum (whiskers). The horizontal dotted line represents FDexp . *** indicate significant
differences (p = 0.001) between FDobs and FDexp according to Wilcoxon signed rank test.

Fig. 4. Non-metric multidimensional scaling (NMDS) showing patterns of functional β -diversity (β -FD) based on the UniFrac metric for transects in different
habitats (stream: circle, terrestrial: triangle, banana plantation: diamond, rice field: square) in the land-use gradient continuous forest (dark gray) – forest
fragments (light gray) – matrix (white). Distances between transects in the two-dimensional NMDS plot represent dissimilarities in β -FD (stress = 0.16).
Transects are grouped according to habitat type.

a distinct turnover in resource-use trait composition between all habitat types indicating that each habitat type contained
a unique set of trait combinations. It is likely that different resource-use trait combinations indicate changes or differences
in resource utilization that affect ecosystem processes. Thus, communities in altered habitats likely provide different, rather
than fewer functions compared to intact forest in our study system, which has recently also been shown for bird assemblages
in the Atlantic forest (De Coster et al., 2015).
All stream habitats, independent of the surrounding habitat type, i.e., continuous forest, forest fragments, or matrix, were
functionally richer than all terrestrial and agricultural habitats. Stream habitats provide a higher structural heterogeneity, as
180 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183

vegetation along streams provides the same strata and microhabitats as the respective terrestrial habitats, and additionally
stream and riparian habitats. Consequently, more niche space is available which fosters a greater dispersion in functional
trait space. Further, species that are not directly stream-dependent may still benefit from the microclimatic conditions
in the vicinity of streams or are at least able to occur there, whereas stream-dependent species (either stream breeding
or semiaquatic) cannot survive without these important landscape elements (Riemann et al., 2015). Beside their general
importance for maintaining high anuran diversity in fragmented landscapes (Ndriantsoa et al., 2017; Riemann et al., 2015;
Rodríguez-Mendoza and Pineda, 2010), streams supported high functional richness and thus resource-use trait diversity in
our study and may therefore stabilize the provisioning of certain ecosystem functions even though these may change within
the habitat alteration gradient.
Species richness was a relevant driver of functional richness regarding amphibian resource-use traits in our system.
Observed differences in functional richness mainly followed differences in species richness, being highest in habitats with
highest species richness (all streams) and lowest in terrestrial matrix habitat (secondary vegetation) where species richness
was lowest. However, species richness was not the only driver of functional richness. We observed a greater loss of functional
richness than species richness suggested (negative deviations from null model expectations) and thus functional clustering
(species similar in functional trait space) in most matrix habitats. This indicates a direct negative effect of human habitat
alterations on functional diversity (Flynn et al., 2009). Nevertheless, functional clustering likewise occurred in terrestrial
continuous forest habitat. Thus, functional clustering might be caused by anthropogenic and natural drivers.
Flynn et al. (2009) proposed that functional clustering may occur if functionally unique species are lost first in
altered habitats. This might have been the case in terrestrial matrix habitat where species loss was highest in our study
system (Ndriantsoa et al., 2017). However, assemblages in terrestrial continuous forest were equally clustered, indicating
that species with unique trait combinations may be scarce in the natural community. Further, assemblages in matrix streams
and banana plantations were likewise clustered, but species richness was not reduced compared to respective habitats in
continuous forest. Thus, functional clustering is not always the result of species loss. Functional clustering is often the result
of environmental trait filtering (HilleRisLambers et al., 2012). Environmental filters may generally be of greater importance
for community assembly in our system than biotic filters like competition (Strauß, et al., 2010). Distinct trait combinations
and a turnover in species composition (Ndriantsoa et al., 2017) indicate that although assemblages were equally clustered,
they pass different environmental filters depending on resource availability in different habitats.
Another idiosyncratic pattern appeared in rice fields. Here, species loss is as severe as in terrestrial matrix habitat (Ndri-
antsoa et al., 2017), but functional richness was as high as in terrestrial continuous forest and assemblages were not
clustered. Mumme et al. (2015) concluded that a greater dissimilarity in functional composition of oil-palm assemblages
indicated that communities are more randomly assembled in highly altered habitats. In contrast, Flynn et al. (2009)
hypothesized that assemblages with low species richness but unique trait composition of species emerge from communities
with high species richness and high redundancy if functionally redundant species are lost first. However, both were not the
case in our system. Since rice fields are structurally homogenous and different from the original forest cover, it is likely that
only species with specific traits can pass these strong environmental filters. Hence, neither species loss nor species assembly
should have been random. Beside very few habitat generalists that also occur in forests, only few open habitat specialists
that differ in their resource-use traits were found in rice fields.
Whereas Ernst et al. (2006) observed different relationships between amphibian species richness and functional richness
in altered habitats in different regions (continents), we found different relationships between changes in functional richness
and species richness in different habitats within the same study system. This underlines that species richness should not be
used as surrogate for functional diversity as patterns of the relationship between species richness and functional diversity
are not only inconsistent within and between taxonomic groups but also at different spatial scales (Flynn et al., 2009; Hevia
et al., 2016).
Direct links between specific ecosystem functions and amphibians are often difficult to measure and to quantify. So far,
only relatively few studies experimentally measured the effects of certain amphibians on specific ecosystem processes
(e.g., Beard et al., 2002) or used real-world local extinctions to compare changes in ecosystem processes and services
between pre and post amphibian declines (e.g., Whiles et al., 2006). Given the magnitude of current amphibian declines,
revealing and quantifying direct links between amphibians and certain ecosystem processes is essential to understand the
real extent of amphibian declines and potential ecosystem consequences that need to be managed (Hocking and Babbitt,
2014). It is especially important to identify traits that are most relevant for certain ecosystem processes and traits that
predispose for habitat alteration sensitivity to ensure the conservation of ecosystem functions. As long as this information
is lacking, trait-based functional approaches that address functional community structures in different habitats with and
without an anthropogenic disturbance background are suitable tools that allow inferences about potential ecosystem
consequences of amphibian declines.
Usually, different sets of species traits produce different functional diversity patterns (Tsianou and Kallimanis, 2016).
Thus, inferences from functional diversity studies are always limited to the traits selected (Petchey and Gaston, 2006). Traits
that determine fragmentation sensitivity may only partially be consistent with or even completely different from resource-
use traits regarded here as relevant for ecosystem functioning. However, our results suggest that species with specific traits
were filtered out in altered habitats which in turn led to assemblages with a specific and partly depauperated set of traits
relevant for ecosystem functioning. Although functional richness was significantly reduced in only one human altered habitat
type compared to respective habitats in continuous forest, distinct changes in trait composition and complex patterns in the
relationship between species richness and functional richness indicate that ecosystem processes like productivity, nutrient
cycling, and energy flows may be affected in human altered landscapes.
 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183 181

5. Conclusion

The ongoing decline and loss of many taxa necessitate that ecosystem consequences of habitat alterations and biodiversity
declines are assessed in more detail and quantified to understand and manage consequences for ecosystem functioning and
ecosystem services (Cardinale et al., 2012). If the aim of conservation and restoration actions is to maintain ecosystem
functioning and stability, integrating functional diversity is required (Cadotte et al., 2011). Habitat alterations caused
complex and idiosyncratic patterns of functional diversity in our study. Our results clearly show that even in systems where
fragmented landscapes still maintain relatively high numbers of species diversity (Ndriantsoa et al., 2017; Riemann et al.,
2015), functional diversity can be altered by habitat fragmentation and conversion. Thus, functional aspects need to be
considered and addressed to a greater extent in conservation planning.

Acknowledgments

We thank the team of Centre ValBio research station for their logistic support, and especially our research assistants A.
Telo and J. Solo for their help in the field. We particularly thank all local landlords and villages who granted permission
to work in their forest fragments and cultivated or fallow land in the matrix. We thank M. Vences, A. Strauss, and M.
Kondermann for helping with species identification via DNA barcoding. J. Nopper provided helpful comments during
the preparation of the manuscript. We thank N.R. Raminosoa and H. Razafindraibe, Mention Zoologie et Biodiversité
Animale, Université d’Antananarivo and are grateful to Madagascar National Parks and the Ministère de l’Environnement,
de l’Ecologie, de la Mer et des Forêts for granting research and export permits (017/033/10/MEF/SG/DGF/DCD.SAP/SLRSE,
03/004/11/MEF/SG/DGF/DCD.SAP/SCB, 045/047/12//MEF/SG/DGF/DCD.SAP/SCB, 115 N-EA06/MG10, 072/079 N-EA06/MG11,
044 N-EA04/MG12). This work was supported by the Deutsche Forschungsgemeinschaft (grant GL 665/1-1 and RO 3064/2-
1). Two anonymous reviewers provided valuable and constructive criticism to a previous version of the manuscript, this is
gratefully acknowledged.

Appendix A. Supplementary data

Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.gecco.2017.03.005.

References

Allnutt, T.F., Ferrier, S., Manion, G., Powell, G.V.N., Ricketts, T.H., Fisher, B.L., Harper, G.J., Irwin, M.E., Kremen, C., Labat, J.-N., Lees, D.C., Pearce, T.A.,
Rakotondrainibe, F., 2008. A method for quantifying biodiversity loss and its application to a 50-year record of deforestation across Madagascar. Conserv.
Lett. 1, 173–181.
Almeida, S.M., Silva, L.C., Cardoso, M.R., Cerqueira, P.V., Juen, L., Santos, M.P.D., 2016. The effects of oil palm plantations on the functional diversity of
Amazonian birds. J. Trop. Ecol. 32, 510–525.
Altig, R., Whiles, M.R., Taylor, C.L., 2007. What do tadpoles really eat? Assessing the trophic status of an understudied and imperiled group of consumers in
freshwater habitats. Freshwat. Biol. 52, 386–395.
Anderson, M.J., 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecol. 26, 32–46.
Beard, K.H., Vogt, K.A., Kulmatiski, A., 2002. Top-down effects of a terrestrial frog on forest nutrient dynamics. Oecologia 133, 583–593.
Benjamini, Y., Hochberg, Y., 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Stat.
Methodol. 57, 289–300.
Butchart, S.H.M., Walpole, M., Collen, B., van Strien, A., Scharlemann, J.P.W., Almond, R.E.A., Baillie, J.E.M., Bomhard, B., Brown, C., Bruno, J., Carpenter, K.E.,
Carr, G.M., Chanson, J., Chenery, A.M., Csirke, J., Davidson, N.C., Dentener, F., Foster, M., Galli, A., Galloway, J.N., Genovesi, P., Gregory, R.D., Hockings, M.,
Kapos, V., Lamarque, J.-F., Leverington, F., Loh, J., McGeoch, M.A., McRae, L., Minasyan, A., Morcillo, M.H., Oldfield, T.E.E., Pauly, D., Quader, S., Revenga,
C., Sauer, J.R., Skolnik, B., Spear, D., Stanwell-Smith, D., Stuart, S.N., Symes, A., Tierney, M., Tyrrell, T.D., Vié, J.-C., Watson, R., 2010. Global biodiversity:
indicators of recent declines. Science 328, 1164–1168.
Cadotte, M.W., Carscadden, K., Mirotchnick, N., 2011. Beyond species: functional diversity and the maintenance of ecological processes and services. J. Appl.
Ecol. 48, 1079–1087.
Cardinale, B.J., Duffy, J., Gonzalez, A., Hooper, D.U., Perrings, C., Venail, P., Narwani, A., Mace, G.M., Tilman, D., A. Wardle, D., Kinzig, A.P., Daily, G.C., Loreau,
M., Grace, J.B., Larigauderie, A., Srivastava, D.S., Naeem, S., 2012. Biodiversity loss and its impact on humanity. Nature 486, 59–67.
Chanson, J.S., Hoffmann, M., Cox, N.A., Stuart, S.N., 2008. The state of the World’s amphibians. In: Stuart, S.N., Hoffmann, M., Chanson, J.S., Cox, N.A., Berridge,
R.J., Ramani, P., Young, B.E. (Eds.), Threatened Amphibians of the World. Lynx Edicions, Barcelona, Spain, IUCN, Gland, Switzerland; Conservation
International, Arlington, Virginia, USA, pp. 33–44.
Cox, N., Berridge, R.J., Church, D., van Dijk, P.P., Kusrini, M., Lau, M., Oldfield, T., Rollins-Smith, L., Xie, F., 2008. Why save amphibians? In: Stuart, S.N.,
Hoffmann, M., Chanson, J.S., Cox, N.A., Berridge, R.J., Ramani, P., Young, B.E. (Eds.), Threatened Amphibians of the World. Lynx Edicions, Barcelona,
Spain, IUCN, Gland, Switzerland; Conservation International, Arlington, Virginia, USA, pp. 23–27.
Davic, R.D., Welsh, H.H., 2004. On the ecological roles of salamanders. Annu. Rev. Ecol. Evol. Syst. 35, 405–434.
De Coster, G., Banks-Leite, C., Metzger, J.P., 2015. Atlantic forest bird communities provide different but not fewer functions after habitat loss. Proc. R. Soc.
B 282, 2014–2844.
182 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183

Dirzo, R., Raven, P.H., 2003. Global state of biodiversity and loss. Ann. Rev. Environ. Resour. 28, 137–167.
Duellman, W.E., Trueb, L., 1994. Biology of Amphibians. The Johns Hopkins University Press, Baltimore.
Ernst, R., Linsenmair, K.E., Rödel, M.-O., 2006. Diversity erosion beyond the species level: Dramatic loss of functional diversity after selective logging in two
tropical amphibian communities. Biol. Cons. 133, 143–155.
Flynn, D.F.B., Gogol-Prokurat, M., Nogeire, T., Molinari, N., Richers, B.T., Lin, B.B., Simpson, N., Mayfield, M.M., Declerck, F., 2009. Loss of functional diversity
under land use intensification across multiple taxa. Ecoll. Lett. 12, 22–33.
Gillespie, G.R., Ahmad, E., Elahan, B., Evans, A., Ancrenaz, M., Goossens, B., Scroggie, M.P., 2012. Conservation of amphibians in Borneo: relative value of
secondary tropical forest and non-forest habitats. Biol. Cons. 152, 136–144.
Glaw, F., Köhler, J., Riva, I.D.L., Vieites, D.R., Vences, M., 2010. Integrative taxonomy of Malagasy treefrogs: combination of molecular genetics, bioacoustics
and comparative morphology reveals twelve additional species of Boophis. Zootaxa 2383, 1–82.
Glaw, F., Thiesmeier, B., 1993. Bioakustische Differenzierung in der Boophis-luteus-Gruppe (Anura: Rhacophoridae), mit Beschreibung einer neuen Art und
einer neuen Unterart. Salamandra 28, 258–269.
Glaw, F., Vences, M., 1994. A Field Guide To the Amphibians and Reptiles of Madagascar, second ed. Vences & Glaw Verlag, Cologne.
Glaw, F., Vences, M., 2007. A Field Guide To the Amphibians and Reptiles of Madagascar, third ed. Vences & Glaw Verlag, Cologne.
Goodman, S.M., Benstead, J.P., 2005. Updated estimates of biotic diversity and endemism for Madagascar. Oryx 39, 73–77.
Grafe, T.U., Stewart, M.M., Lampert, K.P., Rödel, M.-O., 2011. Putting toe clipping into perspective: a viable method for marking anurans. J. Herpetol. 45,
28–35.
Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A.,
Egorov, A., Chini, L., Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of 21st-century forest cover change. Science 342, 850–853.
Harper, G.J., Steininger, M.K., Tucker, C.J., Juhn, D., Hawkins, F., 2007. Fifty years of deforestation and forest fragmentation in Madagascar. Environ. Conserv.
34, 325–333.
Hevia, V., Carmona, C.P., Azcárate, F.M., Torralba, M., Alcorlo, P., Ariño, R., Lozano, J., Castro-Cobo, S., González, J.A., 2016. Effects of land use on taxonomic
and functional diversity: a cross-taxon analysis in a Mediterranean landscape. Oecologia 181, 959–970.
HilleRisLambers, J., Adler, P., Harpole, W., Levine, J., Mayfield, M., 2012. Rethinking community assembly through the lens of coexistence theory. Annu. Rev.
Ecol. Evol. Syst. 43, 227–248.
Hocking, D.J., Babbitt, K.J., 2014. Amphibian contributions to ecosystem services. Herpetol. Conserv. Biol. 9, 1–17.
Irwin, M.T., Wright, P.C., Birkinshaw, C., Fisher, B.L., Gardner, C.J., Glos, J., Goodman, S.M., Loiselle, P., Rabeson, P., Raharison, J.-L., Raherilalao, M.J.,
Rakotondravony, D., Raselimanana, A., Ratsimbazafy, J., Sparks, J.S., Wilmé, L., Ganzhorn, J.U., 2010. Patterns of species change in anthropogenically
disturbed forests of Madagascar. Biol. Cons. 143, 2351–2362.
Kari, S., Korhonen-Kurki, K., 2013. Framing local outcomes of biodiversity conservation through ecosystem services: A case study from Ranomafana,
Madagascar. Ecosys. Service 3, e32–e39.
Kembel, S.W., Cowan, P.D., Helmus, M.R., Cornwell, W.K., Morlon, H., Ackerly, D.D., Blomberg, S.P., Webb, C.O., 2010. Picante: R tools for integrating
phylogenies and ecology. Bioinformatics 26, 1463–1464.
Laurance, W.F., Sayer, J., Cassman, K.G., 2014. Agricultural expansion and its impacts on tropical nature. Trends Ecol. Evol. 29, 107–116.
Lozupone, C., Knight, R., 2005. UniFrac: a new phylogenetic method for comparing microbial communities. Appl. Environ. Microbiol. 71, 8228–8235.
Maechler, M., Rousseeuw, P., Struyf, A., Hubert, M., Hornik, K., 2015. Cluster: Cluster analysis basics and extensions. R package version 2.0.3. URL:
http://CRAN.R-project.org/package=cluster.
Marsh, D.M., Haywood, L.M.B., 2010. Area-based surveys. In: Dodd Jr, C.K. (Ed.), Amphibian Ecology and Conservation –A Handbook of Techniques. Oxford
University Press, Oxford, pp. 247–262.
Mendenhall, C.D., Frishkoff, L.O., Santos-Barrera, G., Pacheco, J., Mesfun, E., Quijano, F.M., Ehrlich, P.R., Ceballos, G., Daily, G.C., Pringle, R.M., 2014. Countryside
biogeography of Neotropical reptiles and amphibians. Ecology 95, 856–870.
Mouchet, M.A., Villéger, S., Mason, N.W.H., Mouillot, D., 2010. Functional diversity measures: an overview of their redundancy and their ability to
discriminate community assembly rules. Funct. Ecol. 24, 867–876.
Mouillot, D., Graham, N.A., Villéger, S., Mason, N.W., Bellwood, D.R., 2013. A functional approach reveals community responses to disturbances. Trends Ecol.
Evol. 28, 167–177.
Mumme, S., Jochum, M., Brose, U., Haneda, N.F., Barnes, A.D., 2015. Functional diversity and stability of litter-invertebrate communities following land-use
change in Sumatra, Indonesia. Biol. Cons. 191, 750–758.
Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000. Biodiversity hotspots for conservation priorities. Nature 403, 853–858.
Ndriantsoa, S.H., Riemann, J.C., Raminosoa, N.R., Rödel, M.-O., Glos, J., 2017. Amphibian diversity in the matrix of a fragmented landscape around Ranomafana
in Madagascar depends on matrix quality. Trop. Conserv. Sci. 10, 1–16.
Nopper, J., Lauströer, B., Rödel, M.-O., Ganzhorn, J.U., 2017. A structurally enriched agricultural landscape maintains high reptile diversity in sub-arid south-
western Madagascar. J. Appl. Ecol 54, 480–488.
Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O’Hara, R.B., Simpson, G.L., Solymos, P., Stevens, M.H.H., Wagner, H., 2016. Vegan: Community
ecology package. R package version 2.3-3. URL: http://CRAN.R-project.org/package=vegan.
Petchey, O.L., Gaston, K.J., 2002. Functional diversity (FD), species richness and community composition. Ecol. Lett. 5, 402–411.
Petchey, O.L., Gaston, K.J., 2006. Functional diversity: back to basics and looking forward. Ecol. Lett. 9, 741–758.
Randrianiaina, R.D., Wollenberg, K.C., Hiobiarilanto, T.R., Strauß, A., Glos, J., Vences, M., 2011. Nidicolous tadpoles rather than direct development in Malagasy
frogs of the genus Gephyromantis. J. Nat. Hist. 45, 2871–2900.
R Core Team, 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R-
project.org/.
Riemann, J.C., Ndriantsoa, S.H., Raminosoa, N.R., Rödel, M.-O., Glos, J., 2015. The value of forest fragments for maintaining amphibian diversity in Madagascar.
Biol. Cons. 191, 707–715.
Rödel, M.-O., Ernst, R., 2004. Measuring and monitoring amphibian diversity in tropical forests. I. An evaluation of methods with recommendations for
standardization. Ecotropica 10, 1–14.
Rodríguez-Mendoza, C., Pineda, E., 2010. Importance of riparian remnants for frog species diversity in a highly fragmented rainforest. Biol. Lett. 6, 781–784.
Strauß, A., Reeve, E., Randrianiaina, R.-D., Vences, M., Glos, J., 2010. The world’s richest tadpole communities show functional redundancy and low functional
diversity: ecological data on Madagascar’s stream-dwelling amphibian larvae. BMC Biol. 10, 12.
Swenson, N.G., 2014. Functional and Phylogenetic Ecology in R. Use R!. Springer, New York.
Tilman, D., 2001. Functional Diversity. In: Levin, S.A. (Ed.), Encyclopedia of Biodiversity. Elsevier, New York, pp. 109–120.
Toledo, L.F., 2005. Predation of juvenile and adult anurans by invertebrates: current knowledge and perspectives. Herpetol. Rev. 36, 395–400.
Toledo, L.F., Ribeiro, R.S., Haddad, C.F.B., 2007. Anurans as prey: an exploratory analysis and size relationships between predators and their prey. J. Zool.
271, 170–177.
Tsianou, M.A., Kallimanis, A.S., 2016. Different species traits produce diverse spatial functional diversity patterns of amphibians. Biodivers. Conserv. 25,
117–132.
 J. Riemann et al. / Global Ecology and Conservation 10 (2017) 173–183 183

Vallan, D., Vences, M., Glaw, F., 2010. Forceps delivery of a new treefrog species of the genus Boophis from eastern Madagascar (Amphibia: Mantellidae).
Amphibia-Reptilia 31, 1–8.
Vences, M., Chiari, Y., Teschke, M., Randrianiaina, R.-D., Raharivololoniaina, L., Bora, P., Vieites, D.R., Glaw, F., 2008. Which frogs are out there? A preliminary
evaluation of survey techniques and identification reliability of Malagasy amphibians. In: Andreone, F. (Ed.), A Conservation Strategy for the Amphibians
of Madagascar. Monografie XLV. Museo Regionale di Scienze Naturali, Torino, pp. 233–252.
Vences, M., Gehara, M., Köhler, J., Glaw, F., 2012. Description of a new Malagasy treefrog (Boophis) occurring syntopically with its sister species, and a plea
for studies on non-allopatric speciation in tropical amphibians. Amphibia-Reptilia 33, 503–520.
Vences, M., Glaw, F., 2002. Two new treefrogs of the Boophis rappiodes group from eastern Madagascar (Amphibia Mantellidae). Trop. Zool. 15, 141–163.
Vences, M., Glaw, F., 2004. Revision of the subgenus Chonomantis (Anura: Mantellidae: Mantidactylus) from Madagascar, with description of two new species.
J. Nat. Hist. 38, 77–118.
Vences, M., Glaw, F., Márquez, R., 2006. The calls of the frogs of Madagascar. In: 3 Audio CDs and Booklet. Alosa-Fonozoo, Barcelona.
Vences, M., Thomas, M., van der Meijden, A., Chiari, Y., Vieites, D.R., 2005. Comparative performance of the 16S rRNA gene in DNA barcoding of amphibians.
Front. Zool. 2, 5.
Vieites, D.R., Wollenberg, K.C., Andreone, F., Köhler, J., Glaw, F., Vences, M., 2009. Vast underestimation of Madagascar’s biodiversity evidenced by an
integrative amphibian inventory. Proc. Natl. Acad. Sci. USA 106, 8267–8272.
Wells, K.D., 2007. The Ecology and Behavior of Amphibians. The University of Chicago Press, Chicago.
Whiles, M.R., Lips, K.R., Pringle, C.M., Kilham, S.S., Bixby, R.J., Brenes, R., Connelly, S., Colon-Gaud, J.C., Hunte-Brown, M., Huryn, A.D., Montgomery, C.,
Peterson, S., 2006. The effects of amphibian population declines on the structure and function of Neotropical stream ecosystems. Front. Ecol. Environ.
4 (1), 27–34.