Habitat structure and host plant specialization drive taxonomic andfunctional composition of Heteroptera in postfire successional habitats

Abstract

Changes in habitat structure are the main driving forces for responses of animal assemblages to fire. According to the disturbance theory, generalist species are expected to outperform specialists in variable environments. Thus, we hypothesized that omnivorous and polyphagous species will become more abundant in unstable postfire successional vegetation, whereas monophagous (specialists), due to their strong dependence on host plants, are expected to respond according to the responses of plant hosts. We compared the responses of true bug (Heteroptera) assemblages in stable (unburnt) versus unstable (postfire successional) environments as this group shows a high diversity of feeding strategies. Redundancy analysis fitted our hypothesis as omnivorous and polyphagous bugs responded positively to fire whereas oligophagous bugs did not. Thus, the most generalized bugs in terms of diet were found in disturbed (burnt) habitats whereas specialized bugs were found in undisturbed (unburnt) habitats. Moreover, the most specialized bugs (monophagous species) responded to fire in accordance to the responses of their specific host plants. Although based on small bipartite networks, the lower modularity in burnt sites corresponded to a scenario of lower segregation of plant resources and fits the higher presence of generalist bugs in these sites. Our results suggest that plant–bug trophic interactions shape the response of Heteroptera to fire, and this response seems to be mediated by the degree of feeding specialization.

Full text

449
http://journals.tubitak.gov.tr/zoology/
Turkish Journal of
Zoology Turk J Zool
(2018) 42: 449-463
© TÜBİTAK
doi:10.3906/zoo-1710-21
Habitat structure and host plant specialization drive taxonomic andfunctional
composition of Heteroptera in postre successional habitats
Eduardo MATEOS*,**

, Marta GOULA

, Teresa SAURAS

, Xavier SANTOS*

Department of Evolutionary Biology, Ecology, and Environmental Sciences, University of Barcelona, Barcelona, Spain
1. Introduction
Fire is a common disturbance in many regions and a
key component to understanding ecosystem functioning
(Bond et al., 2005). e number and extent of wildres
have increased in recent decades, and this pattern is
expected to continue due to current warming (Moriondo
et al., 2006) and land abandonment (Chergui et al., 2017).
For this reason, it is of great interest to analyze the eects
of re on biodiversity and to determine which mechanisms
explain the response of organisms to this disturbance.
Several studies have argued that re can be expected to
have direct and indirect eects on animal communities
(Warren et al., 1987) and to act as an evolutionary driver
for animal diversity (Pausas and Parr, 2018). Although
taxon-dependent, the response of animals to re is
strongly driven by vegetation structure and composition
(Briani et al., 2004; Swan et al., 2015; Santos et al., 2016).
Also, the eect of re on owering phenology could
have a direct (by the absence of owers) or indirect (by
the scarcity of nectar in the early postre years) eect on
pollinator populations (Ne’eman et al., 2000). Accordingly,
animal communities change from unburnt forests to burnt
open habitats in a wide number of taxa following habitat
changes over postre succession (e.g., Brotons et al., 2008
for birds; Driscoll and Henderson, 2008 for reptiles; Santos
et al., 2009 for snails). us, species enter a community
when their preferred habitat type has developed and then
decline as the plant succession proceeds beyond their
optimal habitat conditions (Fox, 1982; Letnic et al., 2004).
In postre scenarios, plants and animals show parallel
trends in response to re related to persistence (ability to
survive), resilience (ability to recover), and mobility traits
(Moretti and Leg, 2009). ese results suggest that plant–
animal interactions may promote postre recolonization,
being a key factor to track ecosystem functioning during
postre succession.
Examining the taxonomic and functional responses
to re is a useful approach to understand the underlying
factors that cause the response (Moretti et al., 2006; Arnan
et al., 2013). is method has illustrated, for example, that
Abstract: Changes in habitat structure are the main driving forces for responses of animal assemblages to re. According to the
disturbance theory, generalist species are expected to outperform specialists in variable environments. us, we hypothesized that
omnivorous and polyphagous species will become more abundant in unstable postre successional vegetation, whereas monophagous
(specialists), due to their strong dependence on host plants, are expected to respond according to the responses of plant hosts. We
compared the responses of true bug (Heteroptera) assemblages in stable (unburnt) versus unstable (postre successional) environments
as this group shows a high diversity of feeding strategies. Redundancy analysis tted our hypothesis as omnivorous and polyphagous
bugs responded positively to re whereas oligophagous bugs did not. us, the most generalized bugs in terms of diet were found in
disturbed (burnt) habitats whereas specialized bugs were found in undisturbed (unburnt) habitats. Moreover, the most specialized bugs
(monophagous species) responded to re in accordance to the responses of their specic host plants. Although based on small bipartite
networks, the lower modularity in burnt sites corresponded to a scenario of lower segregation of plant resources and ts the higher
presence of generalist bugs in these sites. Our results suggest that plant–bug trophic interactions shape the response of Heteroptera to
re, and this response seems to be mediated by the degree of feeding specialization.
Key words: Bipartite network, multivariate ordination, disturbance theory, re, species interaction, trophic specialization
Received: 26.10.2017 Accepted/Published Online: 28.03.2018 Final Version: 26.07.2018
Research Article
* ese authors contributed equally to this work.
** Correspondence: emateos@ub.edu

450
MATEOS et al. / Turk J Zool
in dry temperate regions (e.g., the Mediterranean basin),
functional stability in re-prone ecosystems is achieved by
the replacement of functionally similar species (Moretti et
al., 2009). However, Arnan et al. (2013) reported that re
promotes higher functional diversity in ant communities,
and Mateos et al. (2011) reported that Hymenoptera
parasitoids may respond positively to re according to
the abundance of species to be parasitized. Despite these
pioneering examples, very few studies have examined
functional responses to re by animal communities
(Moretti et al., 2009; Arnan et al., 2013; Santos and Cheylan,
2013; Smith, 2018), and general predictive models need
more empirical evidence.
Animal groups showing a high functional diversity are
adequate model groups to examine functional responses
to re and the role of animal–plant interactions in driving
these responses. To address this question, we have used
Heteroptera (true bugs) since it represents the largest
and most diverse group of hemimetabolous insects
(Schuh and Slater, 1995). Heteroptera species occupy
an enormous array of dierent habitats, performing
a range of ecological functions and services (Henry,
2009). Heteroptera exploit a large range of food sources.
Zoophagy is widespread among terrestrial true bugs, but
the majority of species are phytophagous, feeding on any
part of a plant (Schuh and Slater, 1995). Bugs show a wide
range of host specicity towards their host plants, ranging
from species feeding on a single plant (monophagous), a
genus, or a family (oligophagous) to those that are highly
polyphagous, feeding on multiple plant families (Carver
et al., 1991). Omnivory (also named zoophytophagy),
or the ability of feeding on plants and animals, is also
found among Heteroptera. Floristic composition and
vegetation structure are the environmental factors that
best explain the biodiversity and distribution patterns of
true bug assemblages (Bröring and Wiegleb, 2005; Frank
and Künzle, 2006; Zurbrügg and Frank, 2006) and feeding
group distributions (Torma and Császár, 2013; Torma et
al., 2014, 2017). In the short term, re reduces vegetation
biomass, simplies habitat structure (Keeley et al., 2012),
and drives changes in faunal composition (Kelly and
Brotons, 2017). In addition, the response of animal species
to re can be modeled by postre management, i.e. the
use of burnt vegetation (Bros et al., 2011). us, re and
postre management are expected to induce taxonomic
and functional responses of Heteroptera communities in
accordance to species-specic bug feeding strategies and
degree of specialization.
Our specic objectives were: i) to study variation
in vegetation (plant composition and structure), bug
species, and feeding group composition among re and
managed postre areas; ii) to seek relationships between
vegetation and Heteroptera composition among areas;
and iii) to examine whether plant–bug interactions shape
Heteroptera species and feeding group abundance in burnt
and unburnt areas.
2. Materials and methods
2.1. Study area
e eld work was conducted in Sant Llorenç del Munt i
l’Obac Natural Park (Barcelona Province, NE Spain, Figure
1a). e climate is subhumid Mediterranean with mean
annual temperature of 12.2 °C and annual rainfall around
600 mm. e natural park is prone to fast-spreading res
during hot, dry summers. Field sampling was done in an
area burned on 10 August 2003 (Figure 1b). e burnt
landscape was composed of a pine reforestation (46.4%
Pinus halepensis Mill. and 25.3% P. ni g r a J.F.Arnold) with
small patches of Holm oak forests (18%), scrublands
(6.9%), and abandoned agricultural lands. Postre timber
removal began soon aer the re, and 2 years later,
most of the area was completely logged. Woody debris
remained on the ground. Aer logging, a subarea was
also subsoiled (breaking up soil 30–46 cm deep) to plant
mainly coniferous (Pinus sp.) stands. e area burnt in
2003 included an area that was previously burnt in 1970
(Figure 1b). Between both res, this area was logged and
grazed, being a scrubland habitat in 2003. e study area
was a heterogeneous landscape mosaic both in terms of
habitat structure and postre management. Logging and
subsoiling have dierent impacts on ecosystem function
and structure, as well as on animal and plant diversity
(Lindenmayer and Noss, 2006; Bros et al., 2011). Moreover,
repeat burn regimes have also been reported to aect
vegetation structure and composition and fauna (Fontaine
et al., 2009). For these reasons, logging, subsoiling, and
repeatedly burnt areas were separately considered in
further analyses.
We dened four dierent sampling sites (Figure 1):
1) Repeat-burnt (R) was the scrubland site burnt in 1970,
later logged and grazed, and burnt again in 2003. At this
site, the most frequent plant species, among others, were
Coris monspeliensis L., Helianthemum oelandicum (L.),
Psoralea bituminosa L., and Cistus albidus L. 2) Logged (L)
was the site only burnt in 2003 with subsequent logging;
the most frequent plant species were Filago pyramidata
L., Brachypodium phoenicoides (L.), Ononis minutissima
L., and Sedum sediforme (Jacq.). 3) Subsoiled (S) was
the site only burnt in 2003 with subsequent logging and
subsoiling; the most frequent plant species were Daphne
gnidium L., Cistus albidus, and young individuals of tree
species Quercus ilex L. and Pinus sp. 4) Unburnt (U) was
the control unburnt site in a pine forest with an understory
dominated by Quercus ilex, Linum tenuifolium L., and
ymus vulgaris L. In each sampling site, ve replicated
plots of 20 × 5 m were selected (Figure 1).

451
MATEOS et al. / Turk J Zool
Site and plot selection was constrained by the location
of the areas with dierent postre practices (see Figure 1).
us, site selection was spatially confounded as replicated
plots within the same treatment were clustered. To reduce
this spatial bias, replicated plots were selected to control
for similar slope orientation and ground lithology and to
nd unburnt control plots with similar vegetation structure
and dominant tree species as burnt plots. Moreover, we
examined preburnt vegetation structure with an aerial
photograph taken in 2001, two years before the re. At each
sampling plot, we counted the number of trees in a 50-m
buer and checked dierences among sites. e average
number of pines in a 50-m buer circle around each
sampling plot did not dier among the Logged, Subsoiled,
and Unburnt sites (Logged site: mean = 94.6 pines; SE 16.4,
range 62–151; Subsoiled site: mean = 66.6 pines; SE 20.7,
range 19–132; Unburnt site: mean = 93.6 pines; SE 22.2,
range 44–147; Kruskal–Wallis test, H =1.58, d.f. = 2, P =
0.45). Accordingly, the unburnt control site had similar
vegetation structure to that at burnt sites prior to the 2003
re in terms of the main element of the habitat structure, i.e.
number of pines. In contrast, the Repeat-burnt site in the
2001 aerial picture did not have pines, probably due to the
combined eects of the re and further grazing.
Figure 1. Geographic location of the study area in NE Spain (inner rectangle) and distribution of
sampled sites. Squares: burnt, logged, and subsoiled site (site S); Circles: burnt and logged site (site
L); Triangles: repeat-burnt site (site R); Stars: unburnt site (site U). e striped area was logged aer
the 2003 re and the squared area was subsoiled. Light and dark gray areas represent surfaces aected
by one (2003) and two (1970 and 2003) res, respectively. e green line is the limit of the natural
park. e black area is a village.

452
MATEOS et al. / Turk J Zool
2.2. Field sampling
Field sampling was conducted in July 2007 within the
limits of the Sant Llorenç del Munt i l’Obac Natural Park,
under permits of the Servei de Biodiversitat i Protecció dels
Animals (Direcció General del Medi Natural i Biodiversitat,
Catalan Government, Spain) and Sant Llorenç del Munt
i l’Obac Natural Park (Diputació de Barcelona, Spain).
Heteroptera were collected by net-sweeping ve random
samples from each plot. Each sample consisted of 20
sweeps while walking at a constant speed along a straight
path. e ve samples from each plot were merged,
forming one sample per plot. e net had a light frame of
40 cm in diameter and was 50 cm deep. Specimens were
preserved in 70% ethanol and classied to the species level,
except nymphs (22% of individuals), which were removed
from statistical analyses. Bug species were assigned to one
of the three feeding groups: omnivorous, zoophagous,
and phytophagous, the latter divided into monophagous,
oligophagous, or polyphagous. Bug species identication
and trophic levels were established from the literature
(Appendix 1).
Vegetation (plant species presence and abundance)
was sampled at the same plots as were Heteroptera, using
three complementary techniques: i) presence was recorded
by identifying all plant species in each plot; ii) abundance
of grass, herbs, and shrub species was quantied for each
plot by counting the species along a linear transect with 40
contact points spaced 0.5 m; iii) abundance of tree species
was assessed counting all tree individuals within each plot.
2.3. Statistical analyses
2.3.1. Spatial autocorrelation among sampled sites
In the sampling design, the four sites (R, L, S, U) form a
categorical variable in which each site is a factor (variable
RLSU). To avoid the spatial autocorrelation eect of plots
within each site, we elaborated the variable “position”,
reecting the relative distance of each plot with respect to
the others. For this we performed a principal coordinate
analysis of a truncated matrix of distances among plots. From
this analysis, principal coordinates of neighbor matrices
(PCNMs) were obtained by eigenvalue decomposition,
and the rst (more explicative) PCNM axis was selected
as the explanatory spatial variable “position” (Borcard
et al., 1992). is statistical procedure was conducted by
SpaceMaker2 soware (Borcard et al., 2004; available at
http://adn.biol.umontreal.ca/~numericalecology/old/
spacemaker.html). e variable “position” was included in
further analyses.
2.3.2. Plant and bug species richness and abundance
We quantied plant and bug species richness and
abundance per sampling plot. Bugs were also separated into
feeding groups, and the abundance of each feeding group
was calculated for each plot. We tested the general eect of
sites (variable RLSU) on species richness and abundance
of plants, bugs, and bug feeding group datasets. We used
generalized linear models (glm) with the assumption of
Poisson errors and a logarithmic link function. In analyses
where residual deviance was higher than residual degrees
of freedom, quasi-Poisson errors distribution was used.
ese analyses were performed using the ‘glm’ function of
the ‘stats’ package in R (R Development Core Team, 2008).
With each dataset two glm analyses were done: A) with
sites (RLSU) as factor variable and position as covariable,
B) with sites (RLSU) as factor variable without covariable.
2.3.3. Vegetation composition
From the overall matrix of presence/absence data of
all plant species, we carried out a principal component
analysis (PCA) to establish the similarity in plant species
composition among plots. e raw data matrix was
transformed in a similarity matrix of plots using the
Hellinger distance index (Legendre and Gallagher, 2001).
e Vegan package in R-language (http://cran.r-project.
org/web/packages/vegan/) was used to compute Hellinger
data transformation (decostand function, method = hell)
and PCA analysis (rda function). e primary outcome of
a PCA is a spatial conguration in which the 20 plots are
represented as points, arranged in such a way that their
distances correspond to their similarities in plant species
composition. X and Y coordinates of each plot represent
the best multivariate estimates of their plant species
composition.
2.3.4. Plant and bug species composition between sites
We tested the general eect of sites (variable RLSU) on plant
species composition (using the plant species presence/
absence dataset), bug species composition (using the
bug species abundance dataset), and bug feeding groups
composition (using the percentage of each feeding group
in each sampling plot dataset) by means of the Hellinger
distance index and PERMANOVA analysis (Anderson,
2001). Previously, abundance data were square-root
transformed to reduce asymmetry in the data distribution.
e Vegan package in R-language was used to compute
Hellinger indexes (decostand function, method = hell) and
PERMANOVA analyses with 999 permutations (adonis
function). With each dataset two glm analyses were done:
A) with sites (RLSU) as factor variable and position as
covariable, B) with sites (RLSU) as factor variable without
covariable.
2.3.5. Contribution of vegetation structure and
composition to bug community assemblage
e contribution of vegetation structure to bug species
and feeding group composition in the area was examined
by a set of redundancy analyses (RDAs). Two datasets
were used as response variables: bug species abundance
in each sampling plot, and percentage of feeding groups
in each sampling plot. Nine environmental variables were
introduced in the model: variable position (to analyze

453
MATEOS et al. / Turk J Zool
the eect of spatial autocorrelation) and eight vegetation
variables (to assess the contribution of vegetation structure),
two of them representing plant species composition (scores
of the rst and second axes of vegetation PCA; i.e. variables
PCA1 and PCA2), and the other six representing plant
strata: grass and herb (G), shrub (S), and tree (T) species
richness (variables Grich, Srich, and Trich, respectively),
and grass and herb, shrub, and tree abundance (variables
Gab, Sab, and Tab, respectively). Position and vegetation
variables were standardized in order to eliminate their
physical dimension before being used together to produce
an RDA ordination. For each variable, we performed
an RDA analysis and a permutation test to assess its
marginal eect on the bug species and feeding group
composition. Automatic selection of environmental
variables in the Vegan R package was used to performed
the best RDA model, and a nal RDA was performed
with them. e Vegan R-language package was used for
vegetation data standardization (decostand function,
method = standardize), RDA analyses (rda function),
and permutation tests (anova function with permutation
options).
2.3.6. Antagonistic plant–bug bipartite networks
We constructed bipartite networks taking into account
potential feeding interactions between bug and plant
species. A bipartite network is a network whose vertices or
nodes can be divided into two disjoint sets, in our case bug
species and plant species, such that every edge connects a
node in the rst set to one in the second set. In this study,
each edge represents a plant–bug interaction. Feeding
interactions were established from bug feeding preferences
reviewed in the literature (Appendix 1). We constructed
four local networks containing the potential interactions
between bug species and the possible food plants found
in the plots. For each network we calculated modularity,
i.e. the extent, relative to a null model network, to which
vertices cluster into community groups. Modularity was
assessed with the Girvan and Newman algorithm (Girvan
and Newman, 2002) using the soware MODULAR
(http://sourceforge.net/projects/programmodular/les/).
is algorithm measures the number of clusters or subsets
of nodes within which the node–node connections are
dense. In bipartite networks this means that groups of
species share resources and segregate from another group
of species. Modularity values reached by this algorithm at
each study site were then compared to 10,000 theoretical
(null) networks following Bascompte et al. (2003). Null
networks were based on a set of randomization of plant–
bug interactions, and MODULAR gives the proportion
of theoretical networks with higher modularity values
than the calculated value. Network plots and modularity
structure were visualized using the bipartite R-language
package (Dormann et al., 2008).
3. Results
3.1. Spatial autocorrelation among sampled sites
In the PCNM analysis performed with the data matrix of
distances among plots (Appendix 2), coordinates of axis
1 (42.4% of total variance explained) and axis 2 (29.6% of
total variance explained) summarized the between-plots
relative position. Coordinate values of axis 1 (i.e. PCNM1)
were used as variable “position” for further analyses.
3.2. Plant and bug species richness and abundance
We recorded 135 species of grass, shrub and trees (Appendix
3). e glm analyses indicated that site eect (variable
RLSU) was signicant for all plant datasets except grass
abundance and grass and shrub species richness (Gab,
Grich, and Srich datasets, T2 tests, Table 1). Adding position
eect as covariable (variable position), the eect of site was
only signicant for tree abundance and richness datasets
(Tab and Trich, T1 tests, Table 1). Position eect (variable
position) was signicant for all plant datasets except grass
and shrub species richness (Grich and Srich datasets, T1
tests, Table 1).
Overall, 736 bug specimens from 36 species and 10
families were found (Table 2). Of these, 22% of bugs were
nymph specimens that could not be classied at species level
and were therefore removed from further analyses. Of the
36 bug species found, 12 species were common (represented
by 10 or more individuals). We found no common species
exclusively from the Unburnt site (Table 2). At burnt sites,
ca. 50% of the species (16 species out of 31) were recorded in
only one of the treatments and were present in low numbers.
e glm analyses indicated that site eect (variable RLSU)
was signicant for bug abundance and species richness
(Hatab and Hatsp datasets, T2 tests, Table 3). Adding
position eect as covariable (variable position), the eect of
site was only signicant for bug abundance (Hetab dataset,
T1 tests, Table 3). Position eect (variable position) was
signicant for both bug abundance and richness (Hetab and
Hetsp datasets, T1 tests, Table 3).
Phytophagous bugs were the commonest group in
overall number of species (78%, 28 out of 36 species)
and adult specimens (75%, 432 out of 574 specimens).
Zoophagous and omnivorous groups were represented by 4
species each (11% each). At the unburnt site, phytophagous
bug species were either polyphagous or oligophagous
(50% each), whereas at burnt sites 72% of species were
polyphagous, 16% oligophagous, and 12% monophagous
(Table 2). e glm analyses indicated that site eect
(variable RLSU) was signicant for all bug feeding groups’
abundance except zoophagous (zoo dataset, T2 tests, Table
3). Adding position eect as covariable, the eect of site
was only signicant for monophagous, omnivorous, and
polyphagous feeding groups (mon, om, and pol datasets,
T1 tests, Table 3). Position eect (variable position) was
not signicant only for oligophagous and zoophagous
bugs (olig and zoo datasets, T1 tests, Table 3).

454
MATEOS et al. / Turk J Zool
e PERMANOVA analyses detected that site
(variable RLSU, T2 test, Table 4) was signicant for the
three datasets analyzed (plant species presence/absence,
bug abundance, and bug feeding groups). Adding position
eect as covariable, the eect of site was not signicant for
the bug feeding groups dataset (variable RLSU, T1 test,
Table 4). Position eect (variable position) was signicant
for the three datasets (variable position, T1 tests, Table 4).
3.3. Vegetation composition
In the PCA performed with plant species presence/
absence data (Appendix 4), axis 1 (14.3% of explained
variance) discriminated between unburnt and burnt plots,
whereas axis 2 (10.2% of explained variance) conformed to
a gradient of burnt plots. Coordinate values of axis 1 and
2 (i.e. PCA1 and PCA2) summarized plant composition in
plots and were used for further analyses.
3.4. Vegetation structure and its contribution to bug
community assemblage
e Pearson correlation between variable position and
the eight vegetation variables used in the RDA analyses
was high for several vegetation abundance and richness
variables (correlation values: 0.49 with Gab, –0.74 with
Sab, –0.69 with Tab, 0.28 with Grich, –0.04 with Srich,
–0.58 with Trich,) and for vegetation composition
variables (correlation values: –0.80 with PCA1 and –0.49
with PCA2).
In the RDAs performed with the bug abundance
dataset as response variable and each environmental
variable individually (marginal analyses, Table 5), position
and plant species composition variables (PCA1 and PCA2)
had a signicant contribution, as well as the three plant
strata abundance variables, i.e. abundance of grass (Gab),
shrubs (Sab), and trees (Tab). Shrub (Srich), grass (Grich),
and tree (Trich) species richness did not have a signicant
eect on bug species composition. Although the position
variable had a highly signicant contribution to the
total inertia, due to its high correlation with the other
signicant variables, it was not included for further RDA
analysis. Automatic selection in the Vegan package selected
environmental variables PCA1, PCA2, tree abundance
(Tab), and shrub abundance (Sab) to be included in the
best RDA model (Table 5). In the RDA biplot obtained
with the best model (Figure 2), axis 1 was signicant and
correlated with all environmental variables, especially
with shrub abundance (Sab) and vegetation composition
Table 1. Results of the glm analyses of plants. Test: T1, glm with variable RLSU as factor variable and variable position as covariable.
T2, glm with variable RLSU as factor variable, without covariable. Dataset (response variable): Gab, grass and herb abundance; Sab,
shrub abundance; Tab, tree abundance; Grich, grass and herb species number; Srich, shrub species number; Trich, tree species number.
Df, degrees of freedom; Dev, deviance; Res. df, residual degrees of freedom; Res. dev, residual deviance; P, probability level (ns, not
signicant; * signicance level ≤ 0.05; ** signicance level ≤ 0.01; *** signicance level ≤ 0.001).
Test Dataset Varia ble Df Dev Res. df Res. dev P
T1 Gab position 1 9.30 18 29.59 0.02252 *
T1 Gab RLSU 3 0.69 15 28.90 0.94269 ns
T1 Sab position 1 22.75 18 24.96 0.00012 ***
T1 Sab RLSU 3 1.12 15 23.84 0.86676 ns
T1 Tab position 1 105.45 18 121.06 <0.00001 ***
T1 Tab RLSU 3 59.87 15 61.18 0.001182 **
T1 Grich position 1 1.301 18 14.5 0.25400 ns
T1 Grich RLSU 3 1.68 15 12.81 0.63980 ns
T1 Srich position 1 0.003 18 2.29 0.95280 ns
T1 Srich RLSU 3 0.07 15 2.22 0.99500 ns
T1 Trich position 1 6.06 18 22.79 0.01375 *
T1 Trich RLSU 3 13.10 15 9.68 0.00442 **
T2 Gab RLSU 3 8.38 16 30.52 0.19310 ns
T2 Sab RLSU 3 20.32 16 27.40 0.00752 **
T2 Tab RLSU 3 164.17 16 62.35 <0.00001 ***
T2 Grich RLSU 3 1.85 16 13.94 0.60320 ns
T2 Srich RLSU 3 0.04 16 2.25 0.99730 ns
T2 Trich RLSU 3 18.75 16 10.10 0.00030 ***

455
MATEOS et al. / Turk J Zool
Table 2.
Abundances of Heteroptera species per site. Each numeric column represents the sum of individuals detected in the ve replicate
plots of each site. Abbreviations of sites: R, Repeat-burnt; L, burnt logged; S, burnt logged and subsoiled; U, control unburnt. Abbreviations
of feeding groups (FG): Mon, Monophagous; Oli, Oligophagous; Om, Omnivorous; Ph, Phytophagous; Pol, Polyphagous; Zoo, Zoophagous.
Family Code Code Species R L S U FG
Alydidae CAMLAT 1 Camptopus lateralis 0100Pol
Berytidae GAMPUN 2 Gampsocoris punctipes 0100Pol
Coreidae LOXDEN 3 Loxocnemis dentator 0010Pol
Lygaeidae BEOMAR 5 Beosus maritimus 0030Pol
Lygaeidae HETART 6 Heterogaster artemisiae 1 0 0 0 Olig
Lygaeidae LYGSAR 7 Lygaeosoma sardeum 0100Pol
Lygaeidae MACFAS 8 Macroplax fasciata 9 18 26 0 Pol
Lygaeidae MELALB 9 Melanocoryphus albomaculatus 1000Pol
Lygaeidae NYSCYM 10 Nysius cymoides 5100Pol
Miridae ADELIN 11 Adelphocoris lineolatus 7150Pol
Miridae CYRGEN 12 Cyrtopeltis geniculata 0300Om
Miridae DERSER 13 Deraeocoris serenus 4380Zoo
Miridae HALMAC 14 Halticus macrocephalus 0010Pol
Miridae HETTIG 15 Heterocapillus tigripes 5 52 99 0 Pol
Miridae LEPANC 16 Lepidargyrus ancorifer 72500Pol
Miridae MACCOS 17 Macrolophus costalis 5 14 5 0 Om
Miridae MACATR 18 Macrotylus atricapillus 0110Pol
Miridae MACBIP 19 Macrotylus bipunctatus 10 0 0 0 Olig
Miridae MAVPAY 20 Macrotylus paykulli 32 12 0 0 Pol
Miridae MIMRUG 21 Mimocoris rugicollis 0001Pol
Miridae MONFIL 22 Monalocoris licis 0120Mon
Miridae ORTSTY 23 Orthotylus stysi 5 3 0 18 Olig
Miridae ORTVIR 24 Orthotylus virescens 0 10 0 0 Om
Miridae PACYEL 25 Pachyxyphus yelamosi 0100Mon
Miridae PHYVAR 26 Phytocoris varipes 50 28 1 7 Om
Pentatomidae SCIHEL 29 Sciocoris helferi 0001Pol
Pentatomidae SCIMAC 30 Sciocoris maculatus 0110Pol
Pentatomidae STALUN 31 Staria lunata 0100Pol
Reduviidae CORPER 32 Coranus pericarti 0100Zoo
Reduviidae PHYCRA 33 Phymata crassipes 0001Zoo
Reduviidae RHICUS 34 Rhinocoris cuspidatus 0110Zoo
Rhopalidae CORHYO 35 Corizus hyoscyami 0100Pol
Rhopalidae LIOHYA 36 Liorhyssus hyalinus 2000Pol
Rhopalidae STIPUN 37 Stictopleurus punctatonervosus 0 1 0 0 Olig
Scutelleridae EURTES 39 Eurygaster testudinaria 0 0 0 1 Olig
Tingidae PHAPAR 40 Phaenotropis parvula 48 11 7 0 Mon
Nymphs - unidentied 27 98 12 25 -
Total abundance 218 291 173 54
Zoophagous (Zoo) 4 5 9 1
Phytophagous (Ph) 132 133 146 21
Omnivorous (Om) 55 55 6 7
Total species richness (adults) 15 25 14 6
Zoophagous (Zoo) 1 3 2 1
Phytophagous (Ph) 12 18 10 4
Omnivorous (Om) 2 4 2 1

456
MATEOS et al. / Turk J Zool
Table 3. Results of the glm analyses of bugs. Test: T1, glm with variable RLSU as factor variable and variable position as covariable.
T2, glm with variable RLSU as factor variable, without covariable. Dataset (response variable): Hetab, Heteroptera abundance; Hetsp,
Heteroptera species number; mon, monophagous Heteroptera abundance; olig, oligophagous Heteroptera abundance; om, omnivorous
Heteroptera abundance; pol, polyphagous Heteroptera abundance; zoo, zoophagous Heteroptera abundance. Df, degrees of freedom;
Dev, deviance; Res. df, residual degrees of freedom; Res. dev, residual deviance; P, probability level (ns, not signicant; * signicance level
≤ 0.05; ** signicance level ≤ 0.01; *** signicance level ≤ 0.001).
Test Dataset Varia ble Df Dev Res. df Res. dev P
T1 Hetab position 1 110.05 18 336.79 0.00267 **
T1 Hetab RLSU 3 136.23 15 200.56 0.01089 *
T1 Hetsp position 1 19.33 18 24.90 0.00005 ***
T1 Hetsp RLSU 3 7.22 15 17.68 0.10760 ns
T1 mon position 1 41.95 18 84.08 0.00031 ***
T1 mon RLSU 3 36.37 15 47.71 0.01053 *
T1 olig position 1 8.43 18 82.75 0.17220 ns
T1 olig RLSU 3 25.65 15 57.10 0.12900 ns
T1 om position 1 34.16 18 89.45 0.00003 ***
T1 om RLSU 3 58.88 15 30.57 <0.00001 ***
T1 pol position 1 70.63 18 343.12 0.01108 *
T1 pol RLSU 3 172.71 15 170.41 0.00125 **
T1 zoo position 1 2.90 18 36.49 0.24390 ns
T1 zoo RLSU 3 5.03 15 31.45 0.50260 ns
T2 Hetab RLSU 3 186.98 16 259.87 0.00720 **
T2 Hetsp RLSU 3 25.94 16 18.29 <0.00001 ***
T2 mon RLSU 3 77.16 16 48.87 0.00001 ***
T2 olig RLSU 3 34.07 16 57.11 0.04566 *
T2 om RLSU 3 87.58 16 36.05 <0.00001 ***
T2 pol RLSU 3 190.72 16 223.04 0.00399 **
T2 zoo RLSU 3 7.52 16 31.88 0.29210 ns
Table 4. Global test values from the PERMANOVA analyses made with three datasets: Plants (vegetation presence/absence data), Bugs
Ab (bug abundance data), and Bugs FG% (bug feeding-group percentages). Test: T1, analyses performed with sites (R, L, S, U) as factor
variable and variable position as covariable; T2, analyses performed with sites (R, L, S, U) as factor variable, without covariable. df,
degrees of freedom; F, pseudo-F value; p, permuted P-value (ns, not signicant; ** signicance level ≤ 0.01; *** signicant level ≤ 0.001).
Variables Test df Plants Bugs Ab Bugs FG%
F P F P F P
Position T1 1 2.6342 0.001*** 6.1810 0.001*** 14.9765 0.001***
RLSU T1 3 1.3817 0.005** 2.0126 0.004** 1.6961 0.117 ns
RLSU T2 3 1.9495 0.001*** 3.9788 0.001*** 7.0156 0.001***

457
MATEOS et al. / Turk J Zool
Table 5. Results of the redundancy analyses and Monte Carlo permutation tests (with 999 permutations) analyzing the contribution of
position and vegetation structure variables (environmental variables) on bug species composition with bug species abundance on each
plot as response variable. Analysis: Marginal, RDA analyses performed with only one environmental variable at a time; Best model, RDA
analysis automatically selected as the best model. Var/Axis: environmental variable or axis selected; a denotes environmental variables
selected for the best model. Eigen, eigenvalues. %Var, percentage of variability explained by the corresponding eigenvalue. F, pseudo-F
value in Monte Carlo permutation test. P, permuted P-value in Monte Carlo permutation test (+ signicance level ≤ 0.1; * signicance
level ≤ 0.05; ** signicance level ≤ 0.01; *** signicance level ≤ 0.001).
Analysis Var/Axis Eigen %Var F P
Total inertia 0.6386 100.00
Marginal Position 0.1450 22.70 5.288 0.002 **
Marginal PCA2 a 0.1125 17.61 3.8479 0.005 **
Marginal Sab a 0.0948 14.85 3.1401 0.005 **
Marginal PCA1 a 0.0897 14.04 2.9401 0.015 *
Marginal Tab a 0.0807 12.63 2.6031 0.020 *
Marginal Gab 0.0654 10.24 2.0530 0.039 *
Marginal Tric h 0.0551 8.63 1.6998 0.085
Marginal Grich 0.0281 4.39 0.8270 0.560
Marginal Srich 0.0244 3.82 0.7144 0.780
Best model inertia 0.2771 43.40 2.8752 0.001 ***
Best model Axis 1 0.1584 24.80 6.5757 0.001 ***
Best model Axis 2 0.0593 9.28 2.4612 0.062 +
Figure 2. Biplot of the best model redundancy analysis of the abundances of bug species in the 20 sampling plots (see best model in Table
5). Vegetation variables: PCA1, PCA2, Sab, Gab, Tab (see Section 2 for explanation of variables). For codes (numbers) of bug species, see
Table 2. Bug species 1, 2, 3, 5, 6, 7, 9, 10, 12, 14, 18, 24, 25, 30, 31, 32, 34, 35, and 37 are located inside the circle situated in the center of
the graph. Arrows pointing from origin to each bug species (numbers) have been omitted for clarity. Abbreviations of the sampling plots:
R1 to R5, Repeat-burnt; L1 to L5, burnt logged; S1 to S5, burnt logged and subsoiled; U1 to U5 control unburnt. Axis 1 (horizontal) was
signicant (permutation test: 199 permutations, F ratio 6.5946, P = 0.005), representing 25.4% of the data variance. Axis 2 (vertical) was
also signicant (9.5% of explained variance; permutation test: 399 permutations, F ratio = 2.4677, P = 0.017).
Axis 1 (24.8%)
−0.5 0.0 0.5 1.0
−0.5 0.0 0.5
Axis 2 (9.3%)
8
11
13
15
16
17
19
20
21
22 23
26
29
33
36
39
40
R1
R2
R3
R4
R5
L1
L2
L3
L4
L5
S1
S2
S3
S4
S5
U1
U2
U3
U4
U5
PCA2
Sab
PCA1
−1.0
Tab

458
MATEOS et al. / Turk J Zool
(PCA2). is axis discriminated between unburnt sites
(positive values) and burnt sites (negative values). Axis 2
was also signicant and correlated with variables PCA1,
tree abundance (Tab), shrub abundance (Sab) (positive
correlation), and PCA2 (negative correlation). is
axis discriminated among burnt sites. Bug species were
projected onto Figure 2 in accordance with their relative
abundance on plots.
In the RDAs performed with the bug feeding groups
dataset as response variable and each environmental
variable individually (marginal analyses, Table 6), position
and plant species composition variables PCA1 and PCA2
were signicant, as well as the four plant strata variables
shrub abundance (Sab), tree abundance (Tab), grass
abundance (Gab), and tree richness (Trich). As in previous
analysis, although the position variable had a highly
signicant contribution to the total inertia, due to its high
correlation with the other signicant variables, it was not
included for further RDA analysis. Automatic selection in
the Vegan package selected environmental variables PCA1,
PCA2, and shrub abundance (Sab) to be included in the
best RDA model (Table 6). In the RDA biplot obtained
with the best model (Figure 3), axis 1 was signicant and
discriminated between unburnt (positive values) and
burnt sites (negative values). Bug feeding group structure
diered in relation to re and postre management.
Using the bug functional groups, RDA demonstrated an
association between burnt sites (R, L, and S) and the most
generalist dietary species, i.e. polyphagous-phytophagous
and omnivorous species. In contrast, oligophagous-
phytophagous species were associated to the unburnt
site. Finally, monophagous-phytophagous species were
associated to burnt sites.
3.5. Antagonistic bug–plant bipartite networks
In local networks from the three burnt sites, modularity
did not dier from the null model values, whereas in the
local network of the unburnt site modularity was higher
than the theoretical null models (modularity 0.703, P
= 0.033). e four modules detected in the unburnt site
network are composed of only one or two bug species
each (Figure 4), indicating a high segregation in resource
consumption.
4. Discussion
e four studied sites showed contrasting plant
assemblages, with sharp dierences in plant species
composition and tree abundances and richness. Analysis
of the 2001 aerial photograph conrmed that the prere
vegetation structure was similar between burnt and
unburnt sites except for the plots aected by multiple res.
In parallel, we found specic and functional responses
of the bug assemblage to this habitat transformation.
Table 6. Results of the redundancy analyses and Monte Carlo permutation tests (with 999 permutations) analyzing the contribution of
position and vegetation structure variables (environmental variables) on bug species composition with bug feeding groups’ abundance
on each plot as response variable. Analysis: Marginal, RDA analyses performed with only one environmental variable at a time; Best
model, RDA analysis automatically selected as the best model. Var/Axis: environmental variable or axis selected; a denotes environmental
variables selected for the best model. Eigen, eigenvalues. %Var, percentage of variability explained by the corresponding eigenvalue. F,
pseudo-F value in Monte Carlo permutation test. P, permuted P-value in Monte Carlo permutation test (* signicance level ≤ 0.05; **
signicance level ≤ 0.01; *** signicance level≤ 0.001).
Analysis Var/Axis Eigen % Var F P
Total inertia 0.3255 100.00
Marginal Position 0.1392 42.71 13.419 0.001 ***
Marginal Sab a 0.1086 33.37 9.0156 0.005 **
Marginal PCA1 a 0.0888 27.28 6.7532 0.005 **
Marginal Tab 0.0739 22.72 5.2905 0.015 *
Marginal PCA2 a 0.0569 17.49 3.8158 0.010 **
Marginal Gab 0.0541 16.63 3.5904 0.010 **
Marginal Trich 0.0439 13.48 2.8049 0.048 *
Marginal Srich 0.0051 1.58 0.2883 0.860
Marginal Grich 0.0027 0.84 0.1525 0.930
Best model inertia 0.1662 51.06 5.5643 0.001 ***
Best model Axis 1 0.1488 45.71 14.9482 0.001 ***
Best model Axis 2 0.0106 3.26 1.0723 0.698

459
MATEOS et al. / Turk J Zool
Our results suggest that habitat changes in plant species
composition and tree canopy can shape taxonomic and
functional changes of the bug assemblage. ese changes
accounted for higher abundance of bugs at burnt than
at unburnt sites, and also for species replacement. is
taxonomic response has been similarly reported in a wide
variety of animal taxa previously examined in the study
area (snails [Santos et al., 2009], Hymenoptera [Mateos et
al., 2011], and reptiles [Santos and Poquet, 2010]).
We acknowledge that our results could be partially
biased by a spatial eect due to site selection limitations. e
inclusion of a spatial variable in the glm models indicated
a loss of signicance of site eect only for the number
of bug species and oligophagous bug abundance. Also,
the inclusion of a spatial variable in the PERMANOVA
models indicated a loss of signicance of site eect only
for bug community structure measured by percentage of
feeding groups. e high correlation coecients obtained
between the spatial variable and all vegetation variables
signicantly aecting bug community structure allow
its exclusion from multivariate RDA analyses. Also, the
similarity of prere vegetation structure in the area has
made us convinced that the spatial eect, despite being
statistically signicant in some cases, has not had a real
eect on our results. us, the nal conclusions were made
not taking into consideration the spatial eect.
Changes in habitat structure are major drivers of
animal responses to re, as occurs worldwide in regions
where re is a major and common disturbance (e.g., Briani
et al., 2004; Parr et al., 2004; Lindenmayer et al., 2008;
Izhaki, 2012; Deák et al., 2014). Our study indicated that
bugs are also sensitive to re and postre conditions. is
conclusion is similar to that reported by Ribes (2004) in
another Mediterranean area (located close to our study
area) as bug species richness and abundance were higher
in burnt maquis plots than in unburnt Pinus halepensis
forest plots.
In our study, polyphagous-phytophagous and
omnivorous bugs were associated with burnt sites, this
result suggesting that dietary generalist bug species (i.e.
polyphagous and omnivorous) responded positively to
perturbations such as re. In contrast, the most specialized
oligophagous-phytophagous bugs were associated with
unburnt sites. is gradient of bug feeding specialization
in burnt and unburnt sites ts the ecological theory
as generalist and specialist species are distributed,
respectively, in variable and stable environments
according to some energetic costs (Richmond et al., 2005).
In stable environments, generalists cannot outperform
specialists due to the inherent extra physiological and
behavioral costs associated with a generalist strategy,
which accommodate multiple prey types or are adapted
−1.0 −0.5 0.0 0.5 1.0 1.5
−1.0 −0.5 0.0 0.5 1.0
Axis 1 (45.71 %)
Axis 2 (3.26 %)
Mon
Oli
Om
Pol
Zoo
R1
R2
R3
R4
R5
L1
L2
L3
L4
L5
S1
S2
S3
S4
S5 U1
U2
U3
U4
U5
Sab
PCA1
PCA2
Figure 3. Biplot of the best model redundancy analysis of the percentage of bugs within each trophic group in the 20 sampling plots
(see best model in Table 6). Environmental variables: PCA1, PCA2, Sab, Gab, Tab, Trich (see Section 2 for explanation of variables).
Arrows pointing from origin to each bug feeding group have been omitted for clarity. Abbreviations of bug feeding groups: Om,
omnivorous; Zoo, zoophagous; Pol, phytophagous-polyphagous; Oli, phytophagous-oligophagous; Mon, phytophagous-monophagous.
Abbreviations of the sampling plots: R1 to R5, Repeat-burnt; L1 to L5, burnt logged; S1 to S5, burnt logged and subsoiled; U1 to U5
control unburnt. Axis 1 (horizontal) was signicant (permutation test: 199 permutations, F-ratio = 15.2342, P = 0.005), representing 46.
7% of the data variance. Axis 2 (vertical) was not signicant (6.3% of explained variance; permutation test: 399 permutations, F-ratio =
2.0471, P = 0.100).

460
MATEOS et al. / Turk J Zool
to variable environments (Levins, 1968). In variable or
unpredictable environments, however, these costs may be
small in comparison to benets of the increased plasticity,
and generalists may gain an advantage (Bergman, 1988).
Hence, ecosystems characterized by abrupt environmental
changes triggered by disturbances would promote
generalist species (Futuyma and Moreno, 1988). e same
conclusion was obtained in a study of plant–bee networks
related to re disturbance in a xeric biome in Argentina
(Peralta et al., 2017). Similar eects have also been reported
HETTIG
MACBIP
PHAPAR
HETART
LEPANC
ORTSTY
MACCOS
DERSER
PHYVAR
ADELIN
MACFAS
LIOHYA
MAVPAY
NYSCYM
dorhir
roso
dorpen
thyvul
helsto
scoang
queile
galluc
cisalb
animal
verspc
medmin
erimul
erycam
spajun
eupexi
eupser
fumeri
pislen
sonasp
sonten
onomat
onomin
lacser
R L
SU
MACFAS
MAVPAY
CAMLAT
CYRGEN
ORTVIR
DERSER
RHICUS
CORPER
PHYVAR
MACCOS
ORTSTY
LEPANC
STALUN
HETTIG
MACATR
PHAPAR
SCIMAC
ADELIN
STIPUN
NYSCYM
CORHYO
GAMPUN
LYGSAR
fumeri
pislen
onomat
onomin
astmon
junoxy
gensco
medmin
medspc
animal
galluc
quecoc
pinspc
queile
helsto
scoang
cenasp
dorhir
dorpen
cisalb
thyvul
roso
erimul
erycam
eupser
hiespc
sonspc
sonten
trimon
lacser
orospc
ajucha
sedsed
BEOMAR
PHYVAR
RHICUS
DERSER
MACCOS
MACATR
HALMAC
MACFAS
ADELIN
HETTIG
PHAPAR
LOXDEN
SCIMAC
medmin
medlup
galpum
animal
quecer
queile
cisalb
cissal
pinspc
fumeri
pislen
erycam
eupser
eupspc
rubulm
verspc
erimul
arbune
dorhir
dorpen
gensco
thyvul
roso
EURTES
MIMRUG
ORTSTY
PHYCRA
PHYVAR
SCIHEL
avebro
bradis
brapho
braret
dacglo
koeval
stio
queile
quecer
animal
medspc
erimul
eupser
thyvul
plants
bugsplants
bugs plants
bugsplants
bugs
Figure 4. Local networks indicating interactions between bug species and their potential food sources. For abbreviations of plant species
(le side of the graph), see Appendix 3. For abbreviations of bug species (right side of the graph), see Table 2. Abbreviations of sites:
R, repeat-burnt sites; L, burnt logged; S, burnt logged and subsoiled; U, control unburnt. Lines in network U dene the four modules
detected in the unburnt site network.

461
MATEOS et al. / Turk J Zool
on plants by Valkó et al. (2018), i.e. generalist species can
better tolerate re.
Although our results apparently support the general
trend of generalist species being favored in disturbed
plots, the association of monophagous-phytophagous
bug species to burnt plots is contradictory to previous
statements. ese monophagous-phytophagous bug
species were Monalocoris licis feeding on fern spores of
Dryopteris lix-mas and Eupteris aquilina (Goula, 1986),
Pachyxyphus yelamosi associated to Cistus clusii and C.
monspeliensis (Ribes and Ribes, 2000), and Phaenotropis
parvula associated to Dorycnium suruticosum (Péricart,
1984). Although these plants were not found in the
studied replicates, they are common in burnt sites: e.g.,
Cistus spp. have active postre germination from the seed
bank (Paula and Pausas, 2008), D. suruticosum resprouts
aer re (Rego et al., 1993), and many ferns survive as
rhizomes are protected underground (Paula et al., 2009).
Fire may create early successional habitats that attract
species that specialize in that kind of habitat (Valentine et
al., 2012). In our study site, monophagous-phytophagous
bugs seem to respond to re according to the response of
their hosts. is conclusion highlights that the response of
organisms to a disturbance may be mediated by a complex
of species interactions and stresses the interest of further
investigating the role of antagonistic interactions as a key
element of community responses to disturbances such as
re.
Dierences in network modularity among unburnt and
burnt sites also support the importance of the specialization
gradient to understand the response of bug species to re.
Lower modularity in burnt sites, i.e. lower segregation of
bugs in the use of plant resources, ts the association of
generalist bugs to burnt sites observed in the RDAs. Two
factors suggest that conclusions based on the comparison of
modularity between sites should be accepted with caution:
1) our networks are small, especially in the unburnt site
due to the low number of species found in unburnt plots,
some of them represented by just one specimen; 2) we
do not know what might be the role of monophagous-
phytophagous bugs in the network modularity. However,
results between network modularity and RDA ordination
were consistent. us, the changes detected in network
modularity values between the unburnt and burnt sites
could be due to a functional replacement of bug species (see
also Mateos et al., 2011 and Santos et al., 2014 for similar
conclusions in other groups examined in the same study
area). is conclusion opens future research to examine
the role of plant–bug interactions in postre ecological
trajectories. Plant–herbivore feeding interactions are key
phenomena for understanding processes that maintain
biodiversity (Novotny et al., 2010). ese interactions
make up antagonistic networks characterized by cohesive
groups of interacting species (Bascompte and Jordano,
2006) and promote compartmentalization through
coevolution of specic defenses and counter-defenses that
generate greater specicity (ompson, 2005). Recent
studies have emphasized that interaction networks show
complex responses to disturbance (Piazzon et al., 2011;
Villa-Galaviz et al., 2012). is complexity evidences the
need for further empirical studies to uncover the role of
plant–animal interactions as a driver of postre animal
responses.
Acknowledgments
We are very grateful to managers of the Sant Llorenç del
Munt i l’Obac Natural Park for their logistic support. We
are grateful to Luís Mata for bug classication; Flávia
Marquitti, Paulo R Guimarães, and Márcio S Araujo for
introducing us to soware to examine network modularity;
and Cristina García for her valuable suggestions on an early
dra of the manuscript. Xavier Santos was supported by a
postdoctoral grant (SFRH/BPD/73176/2010) by Fundação
para a Ciência e a Tecnologia, Portugal. Two anonymous
reviewers provided helpful comments that improved the
manuscript.
References
Anderson MJ (2001). A new method for non-parametric multivariate
analysis of variance. Austral Ecol 26: 32-46.
Arnan X, Cerdá X, Rodrigo A, Retana J (2013). Response of ant
functional composition to re. Ecography 36: 1182-1192.
Bascompte J, Jordano P (2006). e structure of plant–animal
mutualistic networks. In: Pascual M, Dunne JA, editors. Food
Webs as Complex Adaptive Networks: Linking Structure to
Dynamics. Oxford, UK: Oxford University Press, pp. 143-159.
Bascompte J, Jordano P, Melián CJ, Olesen JM (2003). e nested
assembly of plant-animal mutualistic networks. P Natl Acad
Sci USA 100: 9383-9387.
Bergman E (1988). Foraging abilities and niche breadths of two
percids, Perca uviatilis and Gymnocephalus cernua, under
dierent environmental conditions. J Anim Ecol 57: 443-453.
Bond WJ, Woodward FI, Midgley GF (2005). e global distribution
of ecosystems in a world without re. New Phytol 165: 525-538.
Borcard D, Legendre P, Avois-Jacquet C, Tuomisto H (2004).
Dissecting the spatial structure of ecological data at multiple
scales. Ecology 85: 1826-1832.
Borcard D, Legendre P, Drapeau P (1992). Partialling out the spatial
component of ecological variation. Ecology 73: 1045-1055.

462
MATEOS et al. / Turk J Zool
Briani DC, Palma ART, Vieira EM, Henriques RPB (2004). Post-re
succession of small mammals in the Cerrado of central Brazil.
Biodiv Conserv 13: 1023-1037.
Bröring U, Wiegleb G (2005). Soil zoology II: Colonization,
distribution, and abundance of terrestrial Heteroptera in
open landscapes of former brown coal mining areas. Ecol Eng
24:135-147.
Bros V, Moreno-Rueda G, Santos X (2011). Does postre management
aect the recovery of Mediterranean communities? e case
study of terrestrial gastropods. For Ecol Manag 261: 611-619.
Brotons L, Herrando S, Pons, P (2008). Wildres and the expansion
of threatened farmland birds: the ortolan bunting Emberiza
hortulana in Mediterranean landscapes. J Appl Ecol 45: 1059-
1066.
Carver M, Gross GF, Woodward, TT (1991). Hemiptera. In:
Naumann ID, Carne PB, Lawrence JF, Nielsen ES, Spradbery
JP, Taylor RW, Whitten MJ, Littlejohn MJ, editors. e Insects
of Australia. 2nd ed. Carlton, Australia: Melbourne University
Press, pp 429-509.
Chergui B, Fahd S, Santos X, Pausas JG (2017). Socioeconomic
factors drive re-regime variability in the Mediterranean
Basin. Ecosystems (in press).
Clarke KR (1993). Non-parametric multivariate analysis of changes
in community structure. Austr J Ecol 18: 117-143.
Deák B, Valkó O, Török P, Végvári Z, Hartel T, Schmotzer A, Kapocsi
I, Tóthmérész B (2014). Grassland res in Hungary? A problem
or a potential alternative management tool? Appl Ecol Env Res
12: 267-283.
Dormann CF, Gruber B, Fründ J (2008). Introducing the bipartite
package: analysing ecological networks. R News 8: 8-11.
Driscoll DA, Henderson MK (2008). How many common reptile
species are re specialists? A replicated natural experiment
highlights the predictive weakness of a re succession model.
Biol Conserv 141: 460-471.
Fontaine JB, Donato DC, Robinson WD, Law BE, Kauman
JB (2009). Bird communities following high-severity re:
Response to single and repeat res in a mixed-evergreen forest,
Oregon, USA. For Ecol Manag 257: 1496-1504.
Fox BJ (1982). Fire and mammalian secondary succession in an
Australian coastal heath. Ecology 63: 1332-1341.
Frank T, Künzle U (2006). Eect of early succession in wildower
areas on bug assemblages (Insecta: Heteroptera). Eur J Entomol
103: 61-70.
Futuyma DJ, Moreno G (1988). e evolution of ecological
specialization. Ann Rev Ecol Syst 19: 207-233.
Girvan M, Newman MEJ (2002). Community structure in social and
biological networks. P Natl Acad Sci USA 99: 7821-7826.
Goula M (1986). Contribución al estudio de los Hemípteros (Insecta,
Heteroptera, Familia Miridae). PhD, Universitat de Barcelona,
Barcelona, Spain (in Spanish).
Henry TJ (2009). Biodiversity of Heteroptera. In: Foottit RG, Adler
PH, editors. Insect Biodiversity: Science and Society. Oxford,
UK: Blackwell Publishing, pp. 223-263.
Izhaki I (2012). e impact of re on vertebrates in the Mediterranean
Basin: an overview. Israel J Ecol Evol 58: 221-233.
Keeley JE, Bond WJ, Bradstock RA, Pausas JG, Rundel PW (2012).
Fire in Mediterranean Ecosystems: Ecology, Evolution and
Management. Cambridge, UK: Cambridge University Press.
Kelly LT, Brotons L, (2017). Using re to promote biodiversity.
Science 355: 1264-1265.
Legendre P, Gallagher ED (2001). Ecologically meaningful
transformations for ordination of species data. Oecologia 129:
271-280.
Letnic M, Dickman CR, Tischler MK, Tamayo B, Beh CL (2004). e
responses of small mammals and lizards to post-re succession
and rainfall in arid Australia. J Arid Environ 59: 85-114.
Levins R (1968). Evolution in Changing Environments. Princeton,
NY, USA: Princeton University Press.
Lindenmayer DB, Noss RF (2006). Salvage logging, ecosystem
processes, and biodiversity conservation. Conserv Biol 20:
949-958.
Lindenmayer DB, Wood JT, MacGregor C, Michael DR, Cunningham
RB, Crane M, Montague-Drake R, Brown D, Muntz R, Driscoll
DA (2008). How predictable are reptile responses to wildre?
Oikos 117: 1086-1097.
Mateos E, Santos X, Pujade-Villar J (2011). Taxonomic and functional
responses to re and post-re management of a Mediterranean
Hymenoptera community. Environ Manag 48: 1000-1012.
Moretti M, de Bello F, Roberts SPM, Potts SG (2009). Taxonomical
vs functional responses of bee communities to re in two
contrasting climatic regions. J Anim Ecol 78: 98-108.
Moretti M, Duelli P, Obrist MK (2006). Biodiversity and resilience
of arthropod communities aer re disturbance in temperate
forests. Oecologia 149: 312-327.
Moretti M, Legg C (2009). Combining plant and animal traits
to assess community functional responses to disturbance.
Ecography 32: 299-309.
Moriondo M, Good P, Durao R, Bindi M, Gianakopoulos C, Corte-
Real J (2006). Potential impact of climate change on re risk in
the Mediterranean area. Clim Res 31: 85-95.
Ne’eman G, Dafni A, Potss SG (2000). e eect of re on ower
visitation rate and fruit set in four core-species in east
Mediterranean scrubland. Plant Ecology 146: 87-104.
Novotny V, Miller SE, Baje L, Balagawi S, Basset Y, Cizek L, Cra KJ,
Dem F, Drew RAI, Hulcr J et al. (2010). Guild-specic patterns
of species richness and host specialization in plant–herbivore
food webs from a tropical forest. J Anim Ecol 79: 1193-1203.
Parr CL, Robertson HG, Biggs HC, Chown SL (2004). Response of
African savanna ants to long-term re regimes. J Appl Ecol 41:
630-642.
Paula S, Arianoutsou M, Kazanis D, Tavsanoglu Ç, Lloret F, Buhk C,
Ojeda F, Luna B, Moreno JM, Rodrigo A, et al. (2009). Fire-
related traits for plant species of the Mediterranean Basin.
Ecology 90: 1420.

463
MATEOS et al. / Turk J Zool
Paula S, Pausas JG (2008). Burning seeds: germinative response to
heat treatments in relation to resprouting ability. J Ecol 96: 543-
552.
Pausas JG, Parr CL (2018). Towards an understanding of the
evolutionary role of re in animals. Evol Ecol (in press).
Peralta G, Stevani EL, Chaco, NP, Dorado J, Vázquez DP (2017).
Fire inuences the structure of plant-bee networks. J Anim
Ecol (in press).
Péricart J (1984). Hémiptères Berytidae euro-méditerranéens. Faune
de France. France et regions limitrophes, 70. Paris, France:
Féderation française des Sociétés de Sciences Naturelles (in
French).
Piazzon M, Larrinaga AR, Santamaría L (2011). Are nested networks
more robust to disturbance? A test using epiphyte-tree,
comensalistic networks. PLoS One 6: e19637.
Rego F, Pereira J, Trabaud L (1993). Modelling community dynamics
of a Quercus coccifera L. garrigue in relation to re using
Markov chains. Ecol Model 66: 251-260.
Ribes E (2004). Els Heteròpters del Parc de Collserola: Faunística,
bioecología i gestió. PhD, Universitat de Barcelona, Barcelona,
Spain (in Catalan).
Ribes E, Ribes J (2000). Noves dades d’Hemípters per a Catalunya
i Territoris limítrofs (Heteroptera). Ses Entom ICHN-SCL 10:
5-29 (in Catalan).
Richmond CE, Breitburg DL, Rose KA (2005). e role of
environmental generalist species in ecosystem function. Ecol
Model 188: 279-295.
Santos X, Badiane A, Matos C (2016). Contrasts in short- and long-
term responses of Mediterranean reptile species to re and
habitat structure. Oecologia 180: 205-2016.
Santos X, Bros V, Miño A (2009). Recolonization of a burnt
Mediterranean area by terrestrial gastropods. Biodiv Conserv
18: 3153-3165.
Santos X, Cheylan M (2013). Taxonomic and functional response of
a Mediterranean reptile assemblage to a repeated re regime.
Biol Conserv 168: 90-98.
Santos X, Mateos E, Bros V, Brotons V, de Mas E, Herraiz JA, Herrando
S, Miño A, Olmo-Vidal JM, Quesada J et al. (2014). Is response
to re inuenced by dietary specialization and mobility? A
comparative study with multiple animal assemblages. PLoS
One 9: e88224.
Santos X, Poquet JM (2010). Ecological succession and habitat
attributes aect the post-re response of a Mediterranean
reptile community. Eu J Wildl Res 56: 895-905.
Schuh RT, Slater JA (1995). True Bugs of the World (Hemiptera,
Heteroptera): Classication and Natural History. Ithaca, NY,
USA: Cornell University Press.
Smith AL (2018). Successional changes in trophic interactions
support a mechanistic model of post-re population dynamics.
Oecologia 186: 129-139.
Swan M, Christie F, Sitters H, York A, Di Stef J (2015). Predicting
faunal re responses in heterogeneous landscapes: the role of
habitat structure. Ecol Appl 25: 2293-2305.
ompson JN (2005). e Geographic Mosaic of Coevolution.
Chicago, IL, USA: University of Chicago Press.
Torma A, Bozsó M, Tölgyesi C, Gallé R (2017). Relationship of
dierent feeding groups of true bugs (Hemiptera: Heteroptera)
with habitat and landscape features in Pannonic salt grasslands.
J Insect Conserv 21: 645-656.
Torma A, Császár P (2013). Species richness and composition
patterns across trophic levels of true bugs (Heteroptera) in the
agricultural landscape of the lower reach of the Tisza River
Basin. J Insect Conserv 17: 35-51.
Torma A, Gallé R, Bozsó M (2014). Eects of habitat and landscape
characteristics on the arthropod assemblages (Araneae,
Orthoptera, Heteroptera) of sand grassland remnants in
Southern Hungary. Agr Ecosyst Environ 196: 42-50.
Valkó O, Kelemen A, Miglécz T, Török P, Deák B, Tóth K, Tóth JP,
Tóthmérész B (2018). Litter removal does not compensate
detrimental re eects on biodiversity in regularly burned
semi-natural grasslands. Sci Total Environ 622-623: 783-789.
Valentine LE, Reaveley A, Johnson B, Fisher R, Wilson BA (2012).
Burning in Banksia Woodlands: how does the re-free period
inuence reptile communities? PLoS One 7: e34448.
Villa-Galaviz E, Boege K, del-Val E (2012). Resilience in plant-
herbivore networks during secondary succession. PLoS One 7:
e53009.
Warren SD, Scifres CJ, Teel PD (1987). Response of grassland
arthropods to burning: a review. Agr Ecosyst Environ 19: 105-
130.
Zurbrügg C, Frank T (2006). Factors inuencing bug diversity
(Insecta: Heteroptera) in semi-natural habitats. Biodiv Conserv
15: 275-294.

1
MATEOS et al. / Turk J Zool
Appendix 1. Identication of bug specimens and
plant–bug interactions were established from the following
literature.
Family Tingidae
Péricart J (1983). Hémiptères Tingidae euro-
méditerranéens. Faune de France. France et régions
limitrophes 69: 1-618 (in French).
Family Miridae
Goula M (1986). Contribución al estudio de los
Hemípteros (Insecta, Heteroptera, Familia Miridae). PhD,
Universitat de Barcelona, Barcelona, Spain (in Spanish).
Wagner E (1974). Die Miridae Hahn, 1931, des
Mittelmeerraumes und der Makaronesischen Inseln
(Hemiptera, Heteroptera). Teil. Entomologische
Abhandlungen herausgegeben vom Staatlichen Museum
für Tierkunde Dresden [Dresden] 39: 1-421 (in German).
Wagner E (1975). Die Miridae Hahn, 1831 des
Mittelmeeraumes und der Makaronesischen Inseln
(Hem.,Het.). Teil 3. Entomologische Abhandlungen
herausgegeben vom Staatlichen Museum für Tierkunde
Dresden [Dresden] 40: 1-483 (in German).
Wheeler AG Jr (2001). Biology of the Plant Bugs
(Hemiptera: Miridae). Pests, Predators, Opportunists.
Ithaca, NY, USA: Cornell University Press.
Family Reduviidae
Putshkov PV, Moulet P (2009). Hémiptères
Reduviidae d’Europe Occidentale. Faune de France.
France et régions limitrophes 92: 1-668 (in French).
Family Berytidae
Péricart J (1984). Hémiptères Berytidae euro-
méditerranéens. Faune de France. France et régions
limitrophes 70: 1-165 (in French).
Family Lygaeidae
Péricart J (1999a). Hémiptères Lygaeidae Euro-
Méditerranéens, 1. Faune de France. France et régions
limitrophes 84A: 1-468 (in French).
Péricart J (1999b). Hémiptères Lygaeidae Euro-
Méditerranéens, 2. Faune de France. France et régions
limitrophes 84B: 1-453 (in French).
Péricart J (1999c). Hémiptères Lygaeidae Euro-
Méditerranéens, 3. Faune de France. France et régions
limitrophes 84C: 1-487 (in French).
Family Alydidae
Moulet P (1995). Hémiptères Coreoidea (Coreidae,
Rhopalidae, Alydidae) Pyrrhocoridae, Stenocephalidae
Euro-Méditerranéens. Faune de France. France et régions
limitrophes
81: 1-336 (in French).
Family Coreidae
Moulet P (1995). Hémiptères Coreoidea (Coreidae,
Rhopalidae, Alydidae) Pyrrhocoridae, Stenocephalidae
Euro-Méditerranéens. Faune de France. France et régions
limitrophes
81: 1-336 (in French).
Family Rhopalidae
Moulet P (1995). Hémiptères Coreoidea (Coreidae,
Rhopalidae, Alydidae) Pyrrhocoridae, Stenocephalidae
Euro-Méditerranéens. Faune de France. France et régions
limitrophes
81: 1-336 (in French).
Family Pentatomidae
Derjanschi VV, Péricart J (2005). Hémiptères
Pentatomoidea euro-méditerranéens 1. Généralités.
Systématique: Première partie. Faune de France 90:
1-494 (in French).
Family Scutelleridae
Ruiz D, Goula M, Inest, E, Monleón T, Pujol M,
Gordún E (2003). Guía de identicación de los chinches de
los cereals (Insecta, Heteroptera) encontrados en los trigos
españoles. Boletín Sanidad Vegetal Plagas 29: 535-552 (in
Spanish).

2
MATEOS et al. / Turk J Zool
Appendix 2. Principal coordinates of neighbor matrices analysis (PCNM) obtained with matrix of distances among plots.
Abbreviations of the sampling plots: R1 to R5, Repeat-burnt; L1 to L5, burnt logged; S1 to S5, burnt logged and subsoiled;
U1 to U5 control unburnt.
PCNM-1 (42.4%)
-6 -4 -2 0 2 4 6 8 10 12
PCNM-2 (29.6%)
-10
-8
-6
-4
-2
0
2
4
6
8
10
R1
R2
R3
R4
R5
L1
L2
L3
L4
L5
S1
S2
S3S4
S5
U1
U2
U3
U4
U5
Appendix 3a

3
MATEOS et al. / Turk J Zool
Appendix 3. Vegetation results. List of plant species (and abbreviations) found at each sampling site and replicate. Each
numeric column includes plant species presence (1) or absence (0) in replicates. Biotype codes (bt code) for each plant
species: G, grass; S, shrub; T, tree. Abbreviations of sites: R, Repeat-burnt; L, burnt logged; S, burnt logged and subsoiled;
U, control unburnt.
bt
code
Repeat burnt Logged Subsoiled Unbu rnt
R1 R2 R3 R4 R5 L1 L2 L3 L4 L5 S1 S2 S3 S4 S5 U1 U2 U3 U4 U5
Ajuga chamaepitys L. ajucha G 10101000100000000010
Allium asphaerocephalon L. allasp G 01010000000000000000
Anacamptis pyram idalis (L.) L.C.M.Richard an apyr G 01000000000000000000
Anagallis arvensis L. anaarv G 10111111101101101000
Andryala integrifolia L.andint G 00000000100000000000
Antirrhinum sp. ant spc G 10100000000100000000
Aphillanthes monspeliensis L. aphmon G 00000110011101011010
Arbutus unedo L. abrune T 00000000001100010000
Argyrolobium zan onii (Turra) P.W.Ball argzan G 11111111111111101110
Asparagus acutifolius L. aspacu G 00000000000000010100
Asperula cynanchica L. aspc yn G 00000001000000000000
Asphodelus cf. aspspc G 01010000000000000000
Asteriscus spinosus (L.) Sch.Bip. astspi G 00000000001000000100
Asterolinon linum-stellatum (L.) Duby astlin G 01000000000000100000
Astragalus monspessulanus L. astmon G 00000110011011111100
Astragalus sesameus L. astses G 100000000 00001000000
Avenula bromoides (Gouan) H.Scholz avebro G 00000001000000001100
Blakstonia perfoliata (L.) Hudson blaper G 01011111010101011000
Brachypodium distachyon (L.) Beauv. bradis G 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0
Brachypodium phoenicoides (L.) Roem. & Schultes brapho G 11101111111111111111
Brachypodium retusun (Pers.) Beauv. braret G 11110100110001101101
Bupleurum fruticescens Loe. ex L. bupfru G 00100000000001000000
Bupleurum fruticosum L. bupfri S 00000100001000000000
Campanula erinus L. cameri G 00001000000000000000

4
MATEOS et al. / Turk J Zool
Carex acca Schreber cara G 00000100111101010010
Carex sp. carspc G 01000000000000000000
Catapodium rigi dum (L.) C.E.Hubb. catrig G 01000011100010100000
Centaurea aspera L. cenasp G 00000010100101000000
Centaurea sp. censpc G 00000000000000000010
Centaurium pulchel lum (Swartz) Druce cenpul G 00000001000001000000
Cistus albidus L. cisalb S 11111101111111101110
Cistus salviifolius L. cissal S 00000000000000100000
Clematis ammula L. clea G 00000000000000010000
Conopodium majus (Gouan) Loret in Loret & Barrandon conmaj G 00000000000000000001
Convolvulus arvensis L. conarv G 10010110101101000100
Convolvulus lineatus L. conlin G 10000000000000000000
Coriaria myrtifolia L. cormyr S 11111110111111111001
Coris monspeliensis L. subsp. monspeliensis cormon G 11111111011111000001
Crupina cf. cruspc G 00000000100000000000
Crupina vulgari s Cass. cruvul G 00100000000000000000
Dactylis glomerata L. dacglo G 01100000001000011110
Daphne gnidi um L. dapgni S 11101010001111110101
Dorycnium hirsutum (L.) Ser. dorhir G 00100100000001000001
Dorycnium pentaphyllum Scop. dor pen G 11111111111111111111
Echium vulgare L. echvul G 00100010001000000000
Epipactis cf. atrorubens epiat r G 00000000000000000011
Erica multiora L. erimul S 11111111011110011111
Erucastrum cf. eruspc G 00100000000000000000
Erucastrum nasturt iifolium (Poir.) O.E.Schulz erunas G 00000000000010000000
Eryngium camp estre L. erycam G 01100011100100011110
Euphorbia cf. exigua eupexi G 01000000000000010000
Euphorbia nicaeensis All. eupnic G 00000000000000000001
Euphorbia serrata L. eupser G 10101010111110000111
Euphorbia sp. eupspc G 00000000000001000000
Festuca gr. ovina fesovi G 00000001000000000000

5
MATEOS et al. / Turk J Zool
Filago pyramidata L. lpyr G 10101111111010100000
Fumana ericoides (Cav.) Gandg. fumeri G 11111111111111111101
Galactites tomentosa Moench ga ltom G 00100000000110100000
Galium lucidum All. galluc G 11000010000000000000
Galium pumilum Murray non Lam. galpum G 00000000001100000000
Genista scorpius (L.) DC. in Lam. & DC. gensco S 00100011111110100010
Globularia alypum L. gloaly S 00011101000000000000
Helianthemum oelandicum (L.) Dum. Cours. heloel G 11111111011110100010
Helichrysum stoechas (L.) Moench helsto G 00100011101000000011
Hieracium sp. hiespc G 00000000011100000010
Hippocrepis comosa L. hipcom G 00000110011111000010
Hypericum perforatum L. hypp er G 00000000000000000010
Hypochoeris radicata L. hypradG 00000001000000000000
Juniperus oxycedrus L. junoxy S 11111111110010011111
Koeleria vallesiana (Honckeny) Gaud. koeval G 00000000000000011000
Lactuca serriola L. lacser G 01100101100111000000
Lavandula latifolia Medic. lavlat G 01011000000000001111
Leontodon taraxacoides (Vill.) Mérat leotar G 11100111101011111000
Leuzea conifera (L.) DC leucon G 01000110010000000000
Ligustrum vulgare L. ligvul T 00000000000000000001
Linum narbonense L. linnar G 00000000000100000000
Linum strictum L. linstr G 10000100010000000000
Linum tenuifolium L. linten G 01000010011110111111
Lonicera implexa Ait. lonimpS 01000010000010101000
Medicago lupulina L. medlup G 00000000000100100000
Medicago minima (l.) L. medminG 01110011100001000000
Medicago s p. medspc G 00000000100000001100
Melilotus sp. melspc G 00000000100000101100
Minuartia cf. minspc G 00000010000000000000
Olea europaea L. oleeur T 00000000000000000100
Onobrychis saxatilis (L.) Lam. onosax G 01000111010010001001

6
MATEOS et al. / Turk J Zool
Ononis minutissima L. onomin S 01111111111111110111
Ononis natrix L. onomat S 11010010100000000000
Orobanche sp. orospc G 00100110111001011001
Oryzopsis miliacea (L.) Asch. & Graebn. orymil G 01000000100000000000
Phagnalon rupestre (L.) D.C. pharup G 10101101000000000000
Phleum phleoides (L.) Karsten phlphl G 00000000000100000000
Pinus sp. pinspc T 00000111011111111111
Pistacia lentiscus L. pislen S 10000100010000111100
Plantago lanceolata L. plalan G 00100001000001001001
Polygala rup estris Pour r. polrup G 01111000100010011100
Populus alba L. popalb T 00000001000000000000
Populus nigra L. popnig T 00000001000000000000
Psoralea bituminosa L. psobit G 11111010100011010011
Quercus cerrioides Willk. & Costa quecer T 00000000001001011011
Quercus coccifera L. quecoc S 00000100000000000000
Quercus ilex L. subsp. ilex queile T 01000111011111111111
Reseda phyteuma L. resphy G 10011000000100100000
Rhamnus alaternus L. rhaala S 00000000000000000100
Rosa sp. rosspac S 00000000001000000000
Rosmarinus ocinalis L. ros o S 11111111111111111111
Rubia peregrina L. subsp. pe regrina rubper G 00101111111111011111
Rubus ulmifolius Schott rubulm S 00000000001111011011
Sanguisorba minor Scop. sanmin G 00101011111100101011
Santolina chamaecyparissus L. sancha G 00000000000000000010
Satureja calamintha (L.) Scheele satcal G 00000000000000000010
Satureja montana L. satmon G 00000000000000011000
Scorzonera angustifolia L. scoang G 01000001000010000010
Sedum sediforme ( Jacq.) Pau sedsed G 01111111110100100000
Sideritis hirsuta L. subsp. hirsuta sidhir G 00000000000101100000
Smilax aspera L. smiasp S 00000000000000001100
Sonchus asper (L.) Hill sonasp G 00100000000000000000

7
MATEOS et al. / Turk J Zool
Sonchus sp. sonspc G 00000000100000000100
Sonchus tenerrimus L. sonten G 11111111101111111100
Spartium junceum L. sp ajun S 10000000000000000000
Staehelina dubia L. stadub G 00000000000000011110
Stipa oneri B reis tr. stio G 00000000110000001110
Silybum marianum (L.) Gaertn. sylmar G 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0
Teucrium botrys L. teubot G 10000000100000100000
Teucrium chamaedrys L. teucha G 00000000100100010000
Teucrium polium L. subsp. capitatum (L.) Arcang. teup ol G 01100000000000101111
ymus vulgaris L. thyvul G 01011111011111011111
Trigonella monspeliaca L. trimon G 00000000100000000000
Ulex parviorus Pour r. ulepar S 00000000000100000000
Urospermum picroides (L.) Scop. ex F.W.Schmidt uropic G 00000000000000000100
Verbascum sp. verspc G 00011000000000100000
Verbena ocinalis L. vero G 10000001000100000000
Vinca diormis Pou rr. vinspc G 00000000000000000001
Viola sp. viospc G 00001000010100000000
Viti s sp. vitspc S 00000010000000000011

8
MATEOS et al. / Turk J Zool
Appendix 4. Principal component analysis (PCA) obtained with the overall matrix of presence/absence data of all plant
species. Abbreviations of the sampling plots: R1 to R5, Repeat-burnt; L1 to L5, burnt logged; S1 to S5, burnt logged and
subsoiled; U1 to U5 control unburnt.
PCA-2 (10.2 %)
−0.6 −0.4 −0.2
0.0
0.2
0.4
0.6
0.8
Appendix
4

U3


R2 U2
R4

L4 U1
U5
R1 R5
R3 U4
S5

L3 S3

L2
L1 S4
L5

S2 S1

−0.5 0.0 0.5
PCA-1 (14.3 %)

Appendix 4. Principal component analysis (PCA) obtained with the overall matrix of
presence/absence data of all plant species. Abbreviations of the sampling plots: R1 to
R5, Repeat-burnt; L1 to L5, burnt logged; S1 to S5, burnt logged and subsoiled; U1 to
U5 control unburnt.