You are on page 1of 9

Forest Ecology and Management 481 (2021) 118704

Contents lists available at ScienceDirect

Forest Ecology and Management


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

Landscape composition is more important than local vegetation structure


for understory birds in cocoa agroforestry systems
Júlia Perez Cabral *, Deborah Faria , José Carlos Morante-Filho *
Applied Conservation Ecology Lab, Programa de Pós-graduação em Ecologia e Conservação da Biodiversidade, Universidade Estadual de Santa Cruz, Rodovia Ilhéus-
Itabuna, km 16, Salobrinho, 45662-000 Ilhéus, Bahia, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Biodiversity-friendly agricultural systems allow the maintenance of native species even in highly fragmented
Agrosystems landscapes by providing corridors to species dispersion and offering supplementary resources for animal pop­
Anthropogenic landscapes ulations. In the tropical region, cocoa agroforestry systems are of great importance for biodiversity conservation
Avifauna
as they maintain part of the native vegetation, and therefore can be used by the local fauna. In this system,
Fragmentation
Habitat loss
understory of native forests is replaced by cocoa trees, which are shaded by large old-growth trees. However, the
Tropical forest persistence of native species in cocoa agroforests depends on local vegetation characteristics but also the land­
scape structure in which these systems are located. Here, we investigated the influence of landscape composition
(i.e. amount of forest cover, cocoa agroforestry and cattle pasture) and local vegetation structure (i.e. number of
native and cocoa trees, basal area of native trees and canopy closure) on understory birds in 18 cocoa agro­
forestry systems located in three regions in the Brazilian Atlantic forest, presenting distinct land use contexts.
Specifically, we assessed the effects of these landscape and local features in predicting richness and abundance
patterns of the entire community, and also in distinct ecological groups, such as forest-dependent and non-forest-
dependent birds, and insectivores, frugivores, and omnivores. Using generalized linear models and Akaike in­
formation criterion, we observed lower species richness of complete community, non-forest and omnivorous
birds in the most deforested region. Also, our findings demonstrated that cocoa agroforests integrated in more
forested landscapes harbor greater richness and abundance of frugivorous birds. Conversely, the increase in
cattle pasture amount at the landscape had a harsh effect on all bird groups evaluated. Regarding local vege­
tation, we observed that the increase of canopy closure leads to greater abundance of insectivorous birds in cocoa
agroforestry systems. Similarly, abundance of non-forest species increased in agroforests with higher number of
cocoa trees. Our study demonstrated that cocoa agroforestry systems can provide complementary habitats for
many species, including forest birds, and therefore can mitigate the effects of habitat loss. However, this key
benefit for bird conservation will be more effective when these agroforestry systems are located in more forested
landscapes, with low amount of cattle pastures. Our findings therefore reinforce the alarming need to maintain
and recover landscape-scale forest amount to ensure species persistence of birds in anthropogenic landscapes,
even in those comprising biodiversity-friendly land uses such as cocoa agroforestry systems.

1. Introduction processes adversely affecting biodiversity, population persistence in


such landscapes is frequently a result of their interaction (Fahrig, 2003;
Natural environments are being converted to human-modified Pardini et al., 2010; Watling et al., 2020). Together, both processes
landscapes (Fahrig, 2003; Curtis et al., 2018), especially in the tropics, shape the anthropogenic landscapes composed of forest fragments of
were climatic conditions favor agriculture and livestock (Malhi et al., different sizes and disturbance levels, usually inserted in the homoge­
2014). Such changes in land use affect the amount and structure of neous matrix dominated by agricultural systems (Malhi et al., 2014;
natural forests that, mainly due to fragmentation and habitat loss, are Morante-Filho et al., 2016). These landscape changes may cause harsh
becoming smaller and more isolated within human-modified landscapes effects on biodiversity, including population declines and increased risk
(Fahrig, 2003). Although habitat loss and fragmentation are different of species extinction, mainly due the decrease in the availability and

* Corresponding authors.
E-mail addresses: juliaperezcabral@gmail.com (J.P. Cabral), deborahuesc@gmail.com (D. Faria), jcmfilho9@hotmail.com (J.C. Morante-Filho).

https://doi.org/10.1016/j.foreco.2020.118704
Received 18 June 2020; Received in revised form 31 August 2020; Accepted 13 October 2020
0378-1127/© 2020 Elsevier B.V. All rights reserved.
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

quality of the remaining habitat (Laurance et al., 1997). Therefore, in Bierregaard, 1995).
highly deforested landscapes, the matrix features (e.g., the matrix types Here, we evaluated the potential contribution of cocoa agroforestry
and nature of predominant matrix) act as major ecological filters, ulti­ systems to maintain understory bird communities in tropical deforested
mately determining species dispersal among fragments and serving of landscapes. Specifically, we investigated the influence of landscape
supplementary habitat in periods of scarce resources (Tscharntke et al., composition (i.e. amount of forest cover, cocoa agroforestry and cattle
2012; Malhi et al., 2014). pasture) and local vegetation structure (i.e. number of native and cocoa
In tropical forest landscapes, the expansion of cattle pastures and trees, basal area of native trees and canopy closure) on understory birds
crop monocultures are the most frequent cause of population decline, in 18 cocoa agroforestry systems located in three regions in the Brazilian
species extinction and alteration of ecological processes (Fahrig, 2003; Atlantic forest, presenting distinct land use contexts. We assessed the
Melo et al., 2013; Laurance et al., 2014). Additionally, forest loss and the effects of landscape and local predictors on bird community richness and
increasing isolation of forest patches hamper species biological flows in abundance, but also in distinct ecological groups, such as forest-
anthropogenic matrices, which may lead to additional stochastic losses dependent and non-forest-dependent birds, and insectivores, frugi­
(Rosenzweig, 1995). Several studies have documented that habitat vores, and omnivores. For this, we tested three hypotheses: (i) richness
generalist species are more likely to use anthropic matrices to maintain and abundance of the complete community would be negatively influ­
their fundamental activities, such as foraging and reproduction sites, enced by the simplification of local vegetation and the decrease of forest
than forest specialist species (Waltert et al., 2005; Bonier et al., 2007; da cover amount at the landscape scale; (ii) richness and abundance of
Silva et al., 2015). Indeed, the high habitat specialization and low forest-dependent birds would be reduced in more intensively managed
vagility from forest-dwelling species often increase their vulnerability to cocoa agroforestry (i.e. presenting a lower number of native tree and
disturbances (Clough et al., 2009). Therefore, matrix features can shape hence a greater canopy openness, and a greater density of cocoa trees)
the dynamics of native species populations in fragmented landscapes and located in deforested landscapes, while non-forest dependent birds
(Fischer et al., 2013; Bohada-Murillo et al., 2020). In this context, would show an opposite response; (iii) trophic guilds would respond
agroforestry systems can perform a key role for maintaining biodiversity differently to changes in local and landscape features. In particular, we
in these landscapes (Perfecto et al., 1996). These agrosystems provide a would expect to observe a drastic decrease on both richness and abun­
more structurally complex habitat compared to land uses devoid of dance of insectivorous birds in highly managed cocoa agroforestry,
arboreal vegetation (Schroth et al., 2004), conferring a greater land­ particularly in farms located in deforested landscapes dominated by
scape permeability to most species, including forest-dwellers. cattle pastures (Stouffer and Bierregaard, 1995; Sekercioglu et al., 2002,
In particular, agroforests are composed of native tree species that 2004). Conversely, we would expect to find an increase of richness and
retain part of the local structure of a forest, such as big trees that increase abundance of non-forest birds, mainly omnivorous species, in these
the shading of the understory layer, and therefore create a more favor­ agroforests (García et al., 2007).
able microclimate, as well as produce a greater amount of resources for
the fauna (Greenberg et al., 2008). These features allow the mainte­ 2. Methods
nance of species diversity (Schroth et al., 2011) and ecological pro­
cesses, including carbon storage (Schroth et al., 2014), being hence 2.1. Study area
considered more biodiversity-friendly than conventional farming sys­
tems (Perfecto and Vandermeer, 2008; Martin et al., 2020). However, This study was conducted in southeastern Bahia State, a region
management intensification of agroforestry through the removal of large originally dominated by Brazilian Atlantic forest. However, land use
trees may reduce the capacity of these systems to maintain biodiversity, changes over the last 40 years have created human-modified landscapes
especially due to the simplification of vegetation structure (Steffan- mainly composed of a mosaic of secondary forests, cocoa (Theobroma
Dewenter et al., 2007; Bohada-Murillo et al., 2020). For instance, several cacao) agroforestry, rubber tree (Hevea brasiliensis) and eucalyptus
studies documented that local intensification of coffee and cocoa agro­ (Eucalyptus sp.) plantations, as well as cattle pastures (Morante-Filho
forestry to increase crop production has drastically reduced native tree et al., 2016). The average annual temperature is 24 ◦ C, and annual
richness and shade levels, putting at risk the animal species that use rainfall averages 2000 mm/yr. Although there is no significant seasonal
these systems (Vandermeer, 2011). Similary, agroforestry systems climatic variation, a rainless period may occur from December to March
located in landscapes with low representation of native remnants are (Thomas et al., 1998).
unable to maintain regional species diversity (Faria et al., 2006, 2007). Using the ArcGIS software and satellite images (QuickBird and
Therefore, environmenal factors acting at multiple scales can be pivotal WorldView from 2011; RapidEye from 2009 to 2010), we created digital
to understand the capacity of agroforestry to maintain species diversity maps with a scale of 1:10 000, which is adequate for identifying land
in human-modified landscapes. cover patches based on the visual inspection of differences in color,
Although many studies have shown that several bird species can texture, shape, location and context. In addition, we also performed an
occur in agroforests (Faria et al., 2006; Van Bael et al., 2008; Cassano intensive field validation effort to check and confirm land use classes,
et al., 2009; Clough et al., 2009), the persistence of their populations mainly cocoa agroforests, combining current Google Earth imagery.
depends on the local characteristics and landscape composition in which Then, we randomly selected 18 agroforestry sites, separated by at least 2
these systems are inserted (Harvey and Villalobos, 2007). Also, bird km from each other, along a gradient of forest cover amount (from
responses may depend on species’ ecological traits, even among those 3.45% to 66.74%), located in three regions presenting different land use
groups of species, such as forest-dependent birds, usually considered to contexts (Fig. 1). The northernmost region is characterized by moderate
be sensitive to anthropogenic disturbance (Clough et al., 2009). Several forest cover (49%), high cocoa agroforestry cover (28%), and low
studies highlighted the importance of dietary niche and trophic level as extension of open areas (8%) [hereafter, high agroforestry cover (HAC)
factors that influence the species sensitivity to disturbance (Bregman region]. The region located in the central portion of our study area
et al., 2014). For instance, specific trophic guilds, such as understory (Fig. 1b) presents high forest cover amount (57%), mainly concentrated
insectivorous birds (Sekercioglu et al., 2002) and large frugivorous birds around the Una Biological Reserve and the Una Wildlife Refuge – two
(Sekercioglu et al., 2004), are likely to be the first groups to disappear in federally protected conservation units that have a total area of 34,804
deforested landscapes. However, the proneness of extinction can vary ha. This region is highly heterogeneous although the matrix is domi­
even among sensitive species. In particular, frugivorous birds show a nated by cocoa agroforestry systems (22%) and rubber trees plantations
greater vagility and ability to use complementary habitats to obtain food (10%) [hereafter, high forest cover (HFC) region]. In contrast, the
(Moran and Catterall, 2014) compared to insectivorous species, which southernmost region is highly deforested, presents only 34% of forest
require specific local characteristics in forest patches (Stouffer and cover, and the matrix is notably more homogeneous, being dominated

2
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

Fig. 1. Sampling sites (black dots) located in southeastern Bahia State, Brazil. In (A), we show a high agroforestry cover (HAC) region, whose matrix is predomi­
nantly formed by cocoa agroforestry systems. In (B) we show the high forest cover (HFC) region, composed especially by forest remnants. In (C), we highlight the low
forest cover (LFC) region, composed of small forest patches surrounding by cattle pastures, and low amount of cocoa agroforestry systems.

3
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

by pastures (60%) for extensive livestock production, eucalyptus plan­ We subsequently classified birds either as forest-dependent or non-
tations (5%), and a small amount of cocoa agroforestry (1%) [hereafter, forest-dependent species (hereafter, forest birds and non-forest birds)
low forest cover (LFC) region] (Fig. 1c). based on Stotz et al., 1996 and Bregman et al., 2014 (Appendix B). Forest
birds are considered highly specialized on forest resources, and included
2.2. Landscape metrics and scales of effect endemic species of the Atlantic forest and those inhabiting forest in­
teriors. Yet, non-forest birds do not depend on forest resources, and
Using a patch-landscape approach, we evaluated response variables comprised those species reported in a variety of habitat types including
within cocoa agroforestry sites, and measured landscape attributes forest edges, open vegetation and anthropogenic areas, such as crop
within a specific radius (buffer) from the center of each focal site plantations and pasture. Also, bird species were grouped according to
(McGarigal and Cushman, 2002; Fahrig, 2013). We used ArcGIS to es­ their trophic guild (i.e., insectivores, frugivores, nectivores, omnivores,
timate the percentage of old-growth and secondary forest (forest cover, and granivores, see Appendix B). Trophic categories reflect the main
hereafter), cocoa agroforestry and cattle pasture around each site; all food source of the species, and each bird species was categorized as
metrics related to the landscape composition. We chose these metrics omnivores if the diet was composed of different food items. These
because previous studies have detected their influence on bird diversty classifications were based on our prior knowledge about the ecology and
in human-modified tropical forest landscapes (Reino et al., 2009; information available in the literature for each species (Wilman et al.,
Morante-Filho et al., 2016; Boesing et al., 2018). 2014). For subsequent analyses, we used the information only of the
Because the effect of landscape variables on biodiversity depends on three most recorded guilds - insectivores, frugivores, and omnivores.
the spatial scale at which predictors are measured (i.e. the so-called
“scale of effect”, Fahrig, 2013; Jackson and Fahrig, 2015), we calcu­ 2.5. Statistical analyses
lated each landscape metric within 9 different-sized buffers drawn from
the center of each site, ranging from 200 to 1000 m radius (Appendix A We used generalized linear models to evaluate how richness and
in Supplementary material). We then used linear models to identify abundance of bird complete communities and different ecological
which landscape size was most appropriate to assess the effect of each groups are influenced by environmental predictors. Our models were
landscape predictor on patterns of bird richness and abundance (see composed by only three predictor variables, which included two envi­
results in Appendix A). ronmental descriptors (one local and one landscape predictor) and study
region as a categorical factor due to contrasting surrounding land-use
2.3. Local vegetation structure contexts observed. We also created non-interactive models due to our
limited sample size. For each response variable, we checked the variance
In the center of each cocoa agroforestry site, we established one 100 inflation factor (VIF) to test the multicollinearity between predictors for
× 25 m plot and recorded all old-growth trees with a diameter at breast each model and then removed those variables presenting VIF ≥ 5 (Zuur
height (DBH) ≥ 10 cm, estimated the basal area (m2/ha) and recorded et al., 2009).
the number of cocoa trees, regardless of height or DBH. We also esti­ We fitted a null model for each response variable to verify if the
mated the percentage of canopy closure inside each plot using five models were better than would be expected by chance. We subsequently
hemispherical photographs (Nikon Coolpix 4300 digital camera equip­ used Akaike Information Criterion corrected for small samples (AICc)
ped with hemispherical fisheye lens [Nikon Corp, Tochigi, Japan]), and Akaike weight (i.e. the normalized relative likelihood of each
taken at 1.3 m from the ground and 20 m apart from each other. The model) to select the best model for each response variable (Anderson,
photos were taken between cocoa individuals, as our objective was to 2008). The model presenting the lowest AIC value was considered the
estimate the shading performed only by canopy trees. These photos were most plausible one, and all models presenting a lower than two unities
analyzed using Gap Light Analyzer software (Frazer et al., 1999), and difference in AIC with similar Akaike weight were considered parsi­
then the mean value of the fives photos were considered as our estimate monious. In this case, we selected the simplest model, i.e. the model with
of canopy closure (in percentage) in each site. Altogether, these vari­ fewer parameters. Yet, we always selected the null model when it was
ables were considered as a proxy of the intensity of local management in among the most plausible models, because we believe that there is no
these cocoa agroforests. Specifically, we expected to obtain in highly simpler model to explain a given pattern than chance.
managed cocoa agroforests, a smaller number of native trees and thinner We also evaluated the fit of the best models using the ratio between
trees, and a greater number of cocoa individuals, which will create a residual deviation and residual degrees of freedom, where values >1
greater canopy openness. indicate overdispersion, i.e. variation in the data is larger than the mean
(Zuur et al., 2009). We further fitted the models that presented over­
2.4. Bird surveys dispersion using the Quasi-poisson family. Furthermore, we evaluated
the spatial autocorrelation in the residuals of the best models based on
We sampled understory birds in two field seasons (January to July the Moran’s I autocorrelation coefficient (Appendix D). All statistical
2019), using the mist-net method (Karr, 1981). Bird sampling was un­ analyses and graphs were carried out in R software version 3.5.3 (R Core
affected by period of survey because our region does not present a Team, 2019) using the packages lme4 (Bates et al., 2015), MASS (Ven­
seasonal climatic variation throughout the year (Thomas et al., 1998). In ables and Ripley, 2002), car (Fox and Weisberg, 2019), mgcv (Wood,
each agroforestry, we used 10 mist-nets with a total of 120 m long (each 2017), nlme (Pinheiro et al., 2020), and AICcmodavg (Mazerolle, 2016).
net with 12 m long, 2.5 m high and 31 mm mesh) to capture birds during
three consecutive days per season. The mist-nets were opened from 3. Results
06:00 a.m. to 04:00p.m., covering the period of activity of the diurnal
birds, and reviewed every 30 min in order to reduce the catch stress. We We captured 625 birds in 18 cocoa agroforestry systems, belonging
avoided sampling on rainy and windy days because such conditions to 24 families and 64 species. The families Thraupidae (10 species),
reduce bird movement and therefore may interfere in the capture suc­ Trochilidae (9 species) and Tyrannidae (8 species) showed the highest
cess. Each captured bird was temporarily marked on one of the right- species richness and Pipridae (n = 173), Turdidae (n = 113) and Tro­
wing primary feathers with non-toxic material to avoid recounting the chilidae (n = 110) presented the highest abundance recorded. The most
same individual during sampling. Then, individuals were identified abundant species captured was Manacus manacus, with 103 individuals,
following the scientific nomenclature of the South American Committee followed by Glaucis hirsutus (n = 88), Turdus leucomelas (n = 68) and
for classification (Remsen et al., 2014). In this study, we obtained a total Euphonia violacea (n = 47). We captured 365 forest birds belonging to 34
sampling effort of 60 h per agroforestry site. species and 260 non-forest birds belonging to 30 species. Additionally,

4
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

insectivorous birds comprised the trophic guild most recorded (22 spe­ Table 1
cies, n = 86), followed by omnivores (15 species, n = 174), and frugi­ Best models (ΔAICc ≤ 2) used to explain the relationship between richness and
vores (13 species, n = 243). abundance of bird ecological groups and several environmental predictors
Our results highlighted several parsimonious models (ΔAICc ≤ 2) for recorded in 18 cocoa agroforestry systems located in the Brazilian Atlantic
explaining the effects of environmental predictors on understory bird forest. When several parsimonious models were indicated, we selected the
simplest model (i.e. the model with fewer parameter - k). The significative
communities (Table 1; Appendix C). Additionally, we did not detect
predictors (P ≤ 0.05) in each model are highlighted in bold (more details in
spatial autocorrelation in the residual of all best models (Appendix D),
Appendix C).
indicating independence among sampling sites. We observed a harsh
Bird groups Response Models k Wi
effect of study region on the diversity of some bird groups, especially in ΔAICc
variables
species most adapted to disturbances (Appendix E). Specifically, we
observed lower species richness of complete community, non-forest and Complete bird Richness Abundance of cocoa 0.00 3 0.21
community trees + Cattle pasture
omnivorous birds in the more deforested region (LFC region). This re­
Cattle pasture 0.71 2 0.15
gion also recorded a lower abundance of non-forest and omnivorous Abundance of trees + 0.84 5 0.14
species (Appendix E). Several variables influenced the richness and Cattle pasture +
abundance of birds in cocoa agroforestry sites, but landscape composi­ Region
Cattle pasture + 1.17 4 0.12
tion had a stronger effect (and significant in several cases; Fig. 2) than
Region
local variables (Table 1). Indeed, cocoa agroforestry located in more Abundance Abundance of cocoa 0.00 5 0.98
forested landscapes presented greater richness and abundance of trees + Cattle pasture
frugivorous birds, with a similar pattern obtained for the abundance of + Region
forest birds. We also found that the increased representation of cattle Forest- Richness Cattle pasture 0.00 2 0.36
pasture at the landscape level had a negative and significant effect on dependent Abundance of cocoa 1.75 3 0.15
richness and abundance of most bird groups evaluated (Fig. 2). In birds trees + Cattle pasture
particular, we observed lower richness and abundance of complete Abundance Canopy closure + 0.00 5 0.67
Forest cover + Region
community and insectivorous species, and lower richness of forest, non-
forest, frugivorous and omnivorous birds in agroforestry located in Non-forest Richness Cattle pasture 0.00 2 0.11
dependent Canopy closure + 0.40 3 0.09
landscapes dominated by cattle pastures. Our results also showed that
birds Cattle pasture
the amount of agroforestry cover present at the landscape level had a Canopy closure 0.51 2 0.09
weak effect on richness and abundance of understory birds, with only a Abundance of cocoa 1.15 3 0.06
significant negative effect on frugivorous species richness and non-forest trees + Cattle pasture
species abundance across studied sites. Region 1.33 3 0.06
Basal area + Cattle 1.41 3 0.06
We finally verified that local vegetation had a slight effect on un­ pasture
derstory birds. In fact, our finding demonstrated that the increase of Abundance of trees + 1.77 3 0.05
canopy closure drives a significant enhancement in the abundance of Cattle pasture
insectivorous birds in these agroforestry systems. Similarly, the abun­ Canopy closure + 1.92 4 0.04
Region
dance of non-forest species increased in agroforests composed by higher
Abundance Abundance of cocoa 0.00 5 1
number of cocoa trees (Fig. 2; Appendix C). trees + Cocoa
agroforestry +
4. Discussion Region

Frugivorous Richness Cattle pasture 0.00 2 0.18


Although several studies have reported that agroforestry systems can birds Forest cover 1.25 2 0.1
help maintaining biodiversity in human-modified landscapes (Perfecto Cocoa agroforestry 1.41 2 0.09
Abundance Abundance of cocoa 0.00 5 0.43
et al., 1996; Faria et al., 2006; Clough et al., 2009), our findings indi­
trees + Forest cover +
cated that the surrounding landscape context is crucial to determine the Region
extent of their contribution for bird conservation. Five observed patterns Abundance of cocoa 0.63 3 0.31
deserve special attention. First, the highly disturbed region, which trees þ Forest cover
presents low amount of both forest cover and agroforestry, and is mostly Insectivorous Richness Abundance of trees + 0.00 3 0.19
dominated by pastures, harbored a more simplified bird community. birds Cattle pasture
Second, forest amount at the landscape scale positively influenced bird Canopy closure + 0.07 3 0.19
Cattle pasture
communities, mainly for those species that depend on certain resources
Cattle pasture 0.29 2 0.17
provided by forest patches, such as forest-dwelling birds and frugivores. Abundance of cocoa 1.77 3 0.08
Third, cocoa agroforests located in landscapes dominated by cattle trees + Cattle pasture
pasture hosted low diversity - richness and abundance - of bird assem­ Abundance Canopy closure + 0.00 3 0.52
blages, regardless of the evaluated ecological group. Fourth, increasing Cattle pasture

representation of shade plantations at landscape hardly buffered the Omnivorous Richness Region 0.00 3 0.12
species decline observed in deforested landscapes. Finally, local vege­ birds Cattle pasture 0.20 2 0.11
Basal area + Cattle 1.15 3 0.07
tation structure showed only a slight effect on understory birds,
pasture
although agroforests presenting a more closed canopy and a greater Abundance of trees + 1.17 4 0.07
number of cocoa trees showed a greater abundance of insectivores and Region
non-forest birds, respectively. We thus discuss possible mechanisms by Basal area + Region 1.61 4 0.06
which the environmental predictors are shaping understory bird com­ Canopy closure + 1.65 3 0.05
Cattle pasture
munities in cocoa agroforestry systems, and suggest some actions for
Abundance Basal area + Cocoa 0.00 5 0.77
bird conservation in these agricultural landscapes. agroforestry + Region
According to our predictions, we observed impoverished bird as­
ΔAICc: difference in AICc between the best model and the ith model; k:
semblages - at community level and for some ecological groups,
parameter number of the model; Wi: AICc weight. Models are ranked by AICc
including non-forest and omnivorous birds - in the highly deforested
values.
region, composed by small and isolated forest patches, commonly

5
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

Fig. 2. Effects of several environmental predictors estimated in the landscape and local scales on richness and abundance of understory birds in 18 cocoa agroforestry
systems. Blue and red boxes represent significant positive and negative effects, respectively. For more details, see Appendix C. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.)

surrounded by a harsh matrix (i.e. cattle pastures), with low amount of bird species richness and abundance in cocoa agroforestry systems
cocoa agroforestry. These landscape features are expected to act as located in landscapes exhibiting greater representation of pastures
important environmental filters selecting bird species with low habitat (Harvey et al., 2006). Our results clearly corroborated this hypothesis,
specificity (Julliard et al., 2006; Ekroos et al., 2010), and therefore indicating a harsh effect of cattle pastures on all studied bird groups.
leading to biotic homogenization. Indeed, only six species were exclu­ Specifically, open matrices, such as pastures, may represent hard bar­
sively recorded in the LFC region (Appendix F), indicating that species riers limiting the movement of birds across the landscape (Boesing et al.,
composition is very similar to other sampled regions. In addition, 2018), also increasing their risk of predation during dispersion (Biz
agroforests inserted in this disturbed region are likely to provide limited et al., 2017) and, thus, preventing the use of different habitats to supply
resource availability for bird species, hence harboring more simplified ecological requirements of species. Moreover, pastures can induce a
communities. Also, dispersal limitation imposed by pasture matrix can more severe edge effect on agroforestry, modifying microclimatic con­
prevent species from reaching resources available in small and already ditions and hence generating a more adverse environment for birds,
scarce forest patches. In fact, several studies documented that forest especially those more sensitive to disturbances such as forest species and
cover amount is a key predictor for structuring bird communities in insectivores. In accordance with the “landscape insurance hypothesis”
shaded plantations, such as cocoa agroforestry systems (Faria et al., (Tscharntke et al., 2012), we suggest that the impoverishment of bird
2006; Clough et al., 2009; Schroth et al., 2011; Tscharntke et al., 2012). assemblages in landscapes dominated by open areas (e.g. pastures) may
Of particular importance, we showed that landscape composition occur due to a reduction in the availability of complementary and/or
was a stronger driver than local vegetation structure for predicting bird supplementary resources. Therefore, these overall rigorous conditions
communities in the studied cocoa agroforests. Although the best models imposed by cattle pastures act as key environmental filters, creating less
included local variables as predictors of bird diversity, in most cases only diverse bird communities in cocoa agroforestry systems integrated in
landscape features exerted a significant effect on richness and abun­ highly disturbed landscapes.
dance of birds (Appendix C). In concordance with our hypothesis, The increasing representation of agroforestry cover in the landscape
agroforests located in more forested landscapes harbor greater richness exerted only a slight, thought negative, effect on the abundance of non-
and abundance of frugivores and forest-dwelling birds. Forest cover has forest species and richness of frugivorous birds. Specifically, while non-
been widely recognized to be a pivotal driver of biodiversity patterns in forest birds are well represented in cocoa agroforests (Faria et al., 2006),
human-modified landscapes, especially because this predictor is posi­ these species are better adapted to use highly disturbed habitats.
tively related to habitat amount and landscape connectivity for a wide Therefore, landscapes dominated by agroforestry systems can reduce the
range of species, particularly forest species (Fahrig, 2003). For instance, availability of suitable habitats to support large populations of these
several studies documented that cocoa agroforests adjacent to forests species. In fact, studies commonly reported a proliferation of non-forest
present higher bird diversity, including frugivorous, because species birds in landscapes presenting high land use intensification, mainly
that use these systems can obtain vital resources for their life cycle, such composed by open agricultural areas (Goulart et al., 2013; Morante-
as food, nesting sites, shelters and reproductive partners, in forest Filho et al., 2015; Kupsch et al., 2019). In addition, the low represen­
remnants (Faria et al., 2006; Clough et al., 2009; Schroth et al., 2011; tation of frugivore species in these landscapes can be associated to
Tscharntke et al., 2012). Also, a previous study performed in the same resource limitation imposed by cocoa agroforests. The fruits of the cocoa
region found higher fruit availability in more forested landscapes, key trees can be consumed and even dispersed by some mammals, but rarely
resources positively affecting the diversity of forest frugivorous birds used as a food item for birds (Clough et al., 2009), especially small-
(Morante-Filho et al., 2018). In fact, as frugivores depend on seasonal bodied frugivores such as understory species. Although native shade
resources, fruit scarcity in specific periods can be compensated by using trees in this system bear fruits that are consumed by birds, tree diversity
multiple habitats in the landscape (Hampe, 2008). More forested land­ – both richness and abundance - is significantly limited compared to
scapes can facilitate species dispersion, mainly for forest birds, to obtain forests (Sambuichi, 2002), particularly in the understory which is
a wide range of resources and foraging areas that are important to entirely replaced by cocoa trees (Clough et al., 2009). Therefore, local
ensure population persistence. vegetation simplification in cocoa agroforestry reduces fruit availability,
Considering that matrix quality highly mediates biodiversity directly impacting frugivorous birds.
response in human-modified landscapes (Sánchez-de-Jesús et al., 2016; Although agroforestry intensification has been associated to local
Boesing et al., 2018; Shoffner et al., 2018), we expected a decrease in species decline of birds (Clough et al., 2009; Harvey and Villalobos,

6
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

2007), we observed that local vegetation features indicating levels of Specifically, Bahia state lost 3,500 ha of Atlantic forest only in the last
management intensification have a limited influence on understory year (Fundação SOS Mata Atlântica, 2020). Therefore, to guarantee the
birds. Specifically, agroforestry systems characterized by many cocoa effectiveness of agroforestry systems for bird conservation, we suggest
trees and presenting more closed canopy showed a higher abundance of that farmers safeguard the existing forest remnants in their properties,
non-forest and insectivorous birds, respectively. In particular, the local but also promote the expansion of forest cover in severely deforested
increasing on the number of cocoa trees is a feature directly associated to landscapes. In addition, in regions devoted to livestock production the
an increasing level of agroforestry intensification, a management prac­ implantation of silvopastoral systems, which combine pastures with
tice that includes the removal of native shade trees to increase the native trees, can contribute to bird conservation, including forest-
density of cocoa trees (Schroth et al., 2011). Therefore, management dwelling species (Mastrangelo and Gavin, 2012). Based on our results,
intensification leads to local vegetation simplification, favoring the we also recommend that large native trees should be maintained in
proliferation of species adapted to open and disturbed environments, cocoa agroforestry to increase local shading, and specifically benefit
such as non-forest birds (Kupsch et al., 2019). Conversely, plantations insectivorous birds, which are most sensitive and functionally important
characterized by more closed canopy, and thus a closer resemblance to species. This is particularly important given the increasing pace and
the original forest structure, can maintain more favorable environ­ extention by which traditional agroforestry systems, that currently host
mental conditions such as milder temperatures and greater availability ~60% of the regional carbon stocks (Schroth et al., 2015), are being
of resources, supporting more disturbance-sensitive species such as converted by thinning or even replaced by “full sun” plantations to
forest-dwellers (Hansbauer et al., 2010). Indeed, 77% of insectivorous enhance local productivity (Schroth et al., 2011).
birds recorded in our study are forest-dependent species (see Appendix
B) – a group commonly recognized for its high sensitivity to habitat CRediT authorship contribution statement
disturbances due to the low vagility and high dietary specificity
(Sekercioglu et al., 2002; Neuschulz et al., 2013). The increase of shade Júlia Perez Cabral: Conceptualization, Data curation, Formal
levels enhances local conditions to host a larger diversity of potential analysis, Investigation, Writing - original draft, Writing - review &
insect prey and foraging space, consequently attracting more members editing. Deborah Faria: Conceptualization, Funding acquisition,
of this functional bird group (Clough et al., 2009). Nevertheless, beyond Writing - review & editing. José Carlos Morante-Filho: Supervision,
structural changes in local vegetation, intensification practices often Conceptualization, Formal analysis, Funding acquisition, Writing - re­
increase crop dependency on chemical inputs including herbicides and view & editing.
broad-spectrum pesticides, practices with pervasive effects on the local
fauna, including bird species (Chang et al., 2018). Although not evalu­
Declaration of Competing Interest
ated here, it is noteworth that consumption of food items constantly
exposed to inseticides and herbicides in agricultural landscapes, such as
The authors declare that they have no known competing financial
seeds, sprayed soils and prey arthropods, can additionally leads to
interests or personal relationships that could have appeared to influence
population declines of many bird species (Mineau and Whiteside, 2013;
the work reported in this paper.
Eng et al., 2017).

5. Conclusions Acknowledgements

Although several studies documented the importance of agroforests We thank the reviewers and editor for all the suggestions and im­
for the biodiversity conservation in agricultural landscapes (Perfecto provements on the manuscript. We thank Alana Cardoso, Albérico
et al., 1996; Faria et al., 2007; Clough et al., 2009), our findings high­ Queiroz, Diego Flores, Fabrine de Almeida, Gabriela Burattini, Ilana
lighted the overwhelming role of landscape features in which these Araújo, Maísa Matuoka, Marcelo Souza and Sueli Damasceno for their
systems are integrated in shaping local bird diversity patterns. Rein­ help in the fieldwork; and local landowners for allowing us to work on
forcing previous studies (Faria et al., 2006), we observed that cocoa their property. We also thank Goetz Schroth and Maíra Benchimol for
agroforestry systems can harbor a high diversity of birds, mostly revision, Sérgio Lopes for helping to map the study area and FAPESB
comprising forest-dwellers, when located in forested landscapes. How­ (N◦ BOL0035/2018) for the financial support. This study was financed in
ever, when landscapes are dominated by cattle pastures, these systems part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Su­
host impoverished bird communities, with lower diversity of both forest perior – Brazil (CAPES – Financing code 001) and the Instituto Nacional
and non-forest species. Interesting, this is the same pattern reported for de Ciência e Tecnologia em Estudos Interdisciplinares e Trans­
bird diversity in remnant forests, in which landscape-scale forest cover disciplinares em Ecologia e Evolução – INCT IN-TREE (CNPq-proc. n.
strongly determined local diversity patterns, i.e. a significant decay in 465767/2014-1, CAPES-proc. n. 23038.000776/2017-54).
local forest bird diversity was observed as deforestation levels increased
(Morante-Filho et al., 2015). Therefore, the persistence of forest- Appendix A. Supplementary material
dwellers in agroforests or forest patches in modified landscapes
strongly relies on a continuous influx of spillovers from surrounding Supplementary data to this article can be found online at https://doi.
native forests (Tscharntke et al., 2008). Altogether, our results reinforce org/10.1016/j.foreco.2020.118704.
the importance of native forest cover to guarantee the maintenance of
species diversity in tropical landscapes (Arroyo-Rodríguez et al., 2020), References
even when landscapes comprise biodiversity-friendly land uses such as
Anderson, D.R., 2008. Model Based Inference in the Life Sciences: A Primer on Evidence.
cocoa agroforests. In addition, these shade plantations have a limited Springer, New York, p. 184.
contribution as bird diversity reservoirs in pasture-dominated land­ Arroyo-Rodríguez, V., Fahrig, L., Tabarelli, M., Watling, J.I., Tischendorf, L.,
scapes, or even when they are the dominant land use. Benchimol, M., Melo, F.P.L., Morante-Filho, J.C., Santos, B.A., Arasa-Gisbert, R.,
Arce-Peña, N., Cervantes-López, M.J., Cudney-Valenzuela, S., Galan-Acedo, C., San-
This is an alarming result considering the current pace of tropical
José, M., Vieira, I.M.G., Slik, J.W.F., Nowakowski, J., Tscharntke, T., 2020.
deforestation - nearly four million hectares of tropical forest are lost Designing optimal human-modified landscapes for forest biodiversity conservation.
annually (Curtis et al., 2018), mainly for the implantation of agricultural Ecol. Lett. 23, 1404–1420.
areas and cattle pastures. Unfortunately, this critical scenario is similar Bates, D., Mächler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models
using lme4. J. Stat. Soft. 67, 1–48.
for the Brazilian Atlantic Forest. During the period of 2018 and 2019, the Biz, M., Cornelius, C., Metzger, J.P., 2017. Matrix type affects movement behavior of a
biome lost 14,500 ha, 27% more than in the last survey (2017–2018). neotropical understory forest bird. Perspect. Ecol. Conser. 15, 10–17.

7
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

Boesing, A.L., Nichols, E., Metzger, J.P., 2018. Land use type, forest cover and forest Martin, D.A., Osen, K., Grass, I., Hölscher, D., Tscharntke, T., Wurz, A., Kreft, H., 2020.
edges modulate avian cross-habitat spillover. J. Appl. Ecol. 55, 1252–1264. Land-use history determines ecosystem services and conservation value in tropical
Bonier, F., Martin, P.R., Wingfield, J.C., 2007. Urban birds have broader environmental agroforestry. Conserv. Lett. e12740.
tolerance. Biol. Lett. 3, 670–673. Mastrangelo, M.E., Gavin, M.C., 2012. Trade-offs between cattle production and bird
Bohada-Murillo, M., Castaño-Villa, G.J., Fontúrbel, F.E., 2020. The effects of forestry and conservation in an agricultural frontier of the Gran Chaco of Argentina. Conserv.
agroforestry plantations on bird diversity: A global synthesis. Land Degrad. Dev. 31, Biol. 26, 1040–1051.
646–654. Mazerolle, M.J., 2016. AICcmodavg: Model selection and multimodel inference based on
Bregman, T.P., Sekercioglu, C.H., Tobias, J.A., 2014. Global patterns and predictors of (Q)AIC(c). R package version 2.0-4.
bird species responses to forest fragmentation: Implications for ecosystem function McGarigal, K., Cushman, S.A., 2002. Comparative evaluation of experimental approaches
and conservation. Biol. Conserv. 169, 372–383. to the study of habitat fragmentation effects. Ecol. Appl. 12, 335–345.
Cassano, C.R., Schroth, G., Faria, D., Delabie, J.H.C., Bede, L., 2009. Landscape and farm Melo, F.P.L., Arroyo-Rodríguez, V., Fahrig, L., Martínez-Ramos, M., Tabarelli, M., 2013.
scale management to enhance biodiversity conservation in the cocoa producing On the hope for biodiversity-friendly tropical landscapes. Trends Ecol. Evol. 28,
region of southern Bahia, Brazil. Biodivers. Conserv. 18, 577–603. 461–468.
Chang, C.H., Karanth, K.K., Robbins, P., 2018. Birds and beans: Comparing avian Mineau, P., Whiteside, M., 2013. Pesticide Acute Toxicity Is a Better Correlate of U.S.
richness and endemism in arabica and robusta agroforests in India’s Western Ghats. Grassland Bird Declines than Agricultural Intensification. PLoS One 8, e57457.
Sci. rep. 8, 1–9. Moran, C., Catterall, C.P., 2014. Responses of seed-dispersing birds to amount of
Clough, Y., Putra, D.D., Pitopang, R., Tscharntke, T., 2009. Local and landscape factors rainforest in the landscape around fragments. Conserv. Biol. 28, 551–560.
determine functional bird diversity in Indonesian cacao agroforestry. Biol. Conserv. Morante-Filho, J.C., Faria, D., Mariano-Neto, E., Rhodes, J., 2015. Birds in
142, 1032–1041. anthropogenics landscapes: the responses of ecological groups to forest loss in the
Curtis, P.G., Slay, C.M., Harris, N.L., Tyukavina, A., Hansen, M.C., 2018. Classifying Brazilian Atlantic Forest. PLoS One 10, e0128923.
drivers of global forest loss. Science 361, 1108–1111. Morante-Filho, J.C., Arroyo-Rodríguez, V., Faria, D., 2016. Patterns and predictors of b-
da Silva, T.W., Dotta, G., Fontana, C.S., 2015. Structure of avian assemblages in diversity in the fragmented Brazilian Atlantic forest: a multiscale analysis of forest
grasslands associated with cattle ranching and soybean agriculture in the Uruguayan specialist and generalist birds. J. Anim. Ecol. 85, 240–250.
savanna ecoregion of Brazil and Uruguay. The Condor 117, 53–63. Morante-Filho, J.C., Arroyo-Rodríguez, V., Pessoa, M.D.S., Cazetta, E., Faria, D., 2018.
Ekroos, J., Heliola, J., Kuussaari, M., 2010. Homogenization of lepidopteran Direct and cascading effects of landscape structure on tropical forest and non-forest
communities in intensively cultivated agricultural landscapes. J. Appl. Ecol. 47, frugivorous birds. Ecol. Appl. 28, 2024–2032.
459–467. Neuschulz, E.L., Brown, M., Farwig, N., 2013. Frequent bird movements across a highly
Eng, M.L., Stutchbury, B.J., Morrissey, C.A., 2017. Imidacloprid and chlorpyrifos fragmented landscape: the role of species traits and forest matrix. Anim. Conserv. 16,
insecticides impair migratory ability in a seed-eating songbird. Sci. Rep. 7, 1–9. 170–179.
Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol. Syst. Pardini, R., de Arruda Bueno, A., Gardner, T.A., Prado, P.I., Metzger, J.P., 2010. Beyond
34, 346–353. the fragmentation threshold hypothesis: regime shifts in biodiversity across
Fahrig, L., 2013. Rethinking patch size and isolation effects: the habitat amount fragmented landscapes. PLoS One 5, e13666.
hypothesis. J. Biogeog. 40, 1649–1663. Perfecto, I., Rice, R.A., Greenberg, R., Van der Voort, M.E., 1996. Shade coffee: a
Faria, D.M., Laps, R.R., Baumgarten, J., Cetra, M., 2006. Bat and bird assemblages from disappearing refuge for biodiversity: shade coffee plantations can contain as much
forest and shade cacao plantations in two contrasting landscapes in the Atlantic biodiversity as forest habitats. BioScience 46, 598–608.
Forest of southern Bahia, Brazil. Biodivers. Conserv. 15, 587–612. Perfecto, I., Vandermeer, J., 2008. Biodiversity conservation in tropical agroecosystems:
Faria, D., Paciencia, M.L.B., Dixo, M., Laps, R.R., Baumgarten, J., 2007. Ferns, frogs, a new conservation paradigm. Ann. NY. Acad. Sci. 1134, 173–200.
lizards, birds and bats in Forest fragments and shade cacao plantations in two Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., R Core Team, 2020. nlme: Linear and
contrasting landscapes in the Atlantic forest. Brazil. Biodivers. Conserv. 16, Nonlinear Mixed Effects Models. R package version 3.1-149.
2335–2357. R Core Team, 2019. R: a language and environment for statistical computing. R
Fischer, J., Abson, D.J., Butsic, V., Chappell, M.J., Ekroos, J., Hanspach, J., Foundation for Statistical Computing, Austria.
Kuemmerle, T., Smith, H.G., Von Wehrden, H., 2013. Land sparing versus land Reino, L., Beja, P., Osborne, P.E., Morgado, R., Fabião, A., Rotenberry, J.T., 2009.
sharing: moving forward. Conserv. Lett. 7, 149–157. Distance to edges, edge contrast and landscape fragmentation: interactions affecting
Fox, J., Weisberg, S., 2019. An R Companion to Applied Regression, Third edition. Sage, farmland birds arround Forest plantations. Biol. Conserv. 142, 824–838.
Thousand Oaks, CA, p. 608. Remsen, J.V., Cadena, C.D., Jaramillo, A., Nores, M., Pacheco, J.F., Pérez-Emen, J., 2014.
Frazer, G.W., Canham, C.D., Lertzman, K.P., 1999. Gap Light Analyzer (GLA), Version A classification of the bird species of. South America American Ornithologists Union.
2.0: imaging software to extract canopy structure and gap light transmission indices Rosenzweig, M.L., 1995. Species diversity in space and time. Cambridge University Press,
from true-colour fish eye photographs. users manual and program documentation. Cambridge, UK, p. 436.
Simon Fraser University, Burnaby, British Columbia, Canada. Sambuichi, R.H.R., 2002. Fitossociologia e diversidade de espécies arbóreas em cabruca
Fundação SOS Mata Atlântica, 2020. Atlas dos remanescentes florestais da Mata (Mata Atlântica raleada sobre plantação de cacau) na região sul da Bahia, Brasil.
Atlântica: período 2018–2019 - Relatório técnico. Fundação SOS Mata Atlântica; Acta Bot. Bras. 16, 89–101.
Instituto Nacional de Pesquisas Espaciais, São Paulo. Sánchez-de-Jesús, H.A., Arroyo-Rodríguez, V., Andresen, E., Escobar, F., 2016. Forest
García, C., Renison, D., Cingolani, A.M., Fernández-Juricic, E., 2007. Avifaunal changes loss and matrix composition are the major drivers shaping dung beetle assemblages
as a consequence of large-scale livestock exclusion in the mountains of Central in a fragmented rainforest. Land. Ecol. 31, 843–854.
Argentina. J. Appl. Ecol. 45, 351–360. Schroth, G., Harvey, C., Vincent, G., 2004. Complex agroforests: their structure,
Goulart, F.F., Salles, P., Saito, C.H., Machado, R.B., 2013. How do different agricultural diversity, and potential role in landscape conservation. Agroforestry and biodiversity
management strategies affect bird communities inhabiting a savanna-forest mosaic? conservation in tropical landscapes. Island Press, Washington, D.C., pp. 227–260
A qualitative reasoning approach. Agr. Ecosyst. Environ. 164, 114–130. Schroth, G., Faria, D., Araujo, M., Bede, L., Van Bael, S.A., Cassano, C.R., Oliveira, L.C.,
Greenberg, R., Perfecto, I., Philpott, S.M., 2008. Agroforests as model systems for tropical Delabie, J.H.C., 2011. Conservation in tropical landscape mosaics: the case of the
ecology. Ecology 913–914. cacao landscape of southern Bahia. Brazil. Biodiver. Conserv. 20, 1635–1654.
Hampe, A., 2008. Fruit tracking, frugivore satiation, and their consequences for seed Schroth, G., Jeusset, A., da Silva Gomes, A., Florence, C.T., Coelho, N.A.P., Faria, D.,
dispersal. Oecologia 156, 137–145. Läderach, P., 2014. Climate friendliness of cocoa agroforests is compatible with
Hansbauer, M.M., Végvári, Z., Storch, I., Borntraeger, R., Hettich, U., Pimentel, R.G., productivity increases. Mitig. Adapt. Strateg. Glob. Change 21, 67–80.
Metzger, J.P., 2010. Microhabitat selection of three forest understory birds in the Schroth, G., Bede, L.C., Paiva, A.O., Cassano, C.R., Amorim, A.M., Faria, D., Mariano-
Brazilian Atlantic rainforest. Biotropica 42, 355–362. Neto, E., Martini, A.M.Z., Sambuichi, R.H.R., Lôbo, R.N., 2015. Contribution of
Harvey, C.A., Medina, A., Sanchez, D.M., Vilchez, S., Hernandez, B., Saenz, J.C., Maes, J. agroforests to landscape carbon storage. Mitig. Adapt. Strat. Glob. Change 20,
M., Casanovaes, F., Sinclair, F.L., 2006. Patterns of animal diversity in different 1175–1190.
forms of tree cover in agricultural landscapes. Ecol. Appl. 16, 1986–1999. Sekercioglu, C.H., Ehrlich, P.R., Daily, G.C., Aygen, D., Goehring, D., Sandi, R.F., 2002.
Harvey, C.A., Villalobos, J.A.G., 2007. Agroforestry systems conserve species-rich but Disappearance of insectivorous birds from tropical forest fragments. Proc. Natl.
modified assemblages of tropical birds and bats. Biodivers. Conserv. 16, 2257–2292. Acad. Sci. 99, 263–267.
Jackson, H.B., Fahrig, L., 2015. Are ecologists conducting research at the optimal scale? Sekercioglu, C.H., Daily, G.C., Ehrlich, P.R., 2004. Ecosystem consequences of bird
Global Ecol. Biogeogr. 24, 52–63. declines. Proc. Natl. Acad. Sci. 101, 18042–18047.
Julliard, R., Clavel, J., Devictor, V., Jiguet, F., Couvet, D., 2006. Spacial segregation os Shoffner, A., Wilson, A.M., Tang, W., Gagné, S.A., 2018. The relative effects of forest
specialists and generalists in bird communities. Ecol. Lett. 9, 1237–1244. amount, forest configuration, and urban matrix quality on forest breeding birds. Sci.
Karr, J.R., 1981. Surveying birds with mist nets. Stud. Avian Biol. 6, 62–67. Rep. 8, 1–12.
Kupsch, D., Vendras, E., Ocampo-Ariza, C., Batáry, P., Motombi, F.N., Bobo, K.S., Steffan-Dewenter, I., Kessler, M., Barkmann, J., Bos, M.M., Buchori, D., Erasmi, S.,
Waltert, M., 2019. High critical forest habitat thresholds of native bird communities Faust, H., Gerold, G., Glenk, K., Gradstein, S.R., Guhardja, E., Harteveld, M.,
in Afrotropical agroforestry landscapes. Biol. Conserv. 230, 20–28. Hertel, D., Höhn, P., Kappas, M., Köhler, S., Leuschner, C., Maertens, M.,
Laurance, W.F., Bierregaard Jr, R.O., Leigh Jr, E.C., 1997. Tropical forest remnants. Marggraf, R., Migge-Kleian, S., Mogea, J., Pitopang, R., Schaefer, M., Schwarze, S.,
Trends Ecol. Evol. 12, 414 414. Sporn, S.G., Steingrebe, A., Tjitrosoedirdjo, S.S., Tjitrosoemito, S., Twele, A.,
Laurance, W.F., Sayer, J., Cassman, K.G., 2014. Agricultural expansion and its impacts on Weber, R., Woltmann, L., Zeller, M., Tscharntke, T., 2007. Tradeoffs between
tropical nature. Trends Ecol. Evol. 29, 107–116. income, biodiversity, and ecosystem functioning during tropical rainforest
Malhi, Y., Gardner, T.A., Goldsmith, G.R., Silman, M.R., Zelazowski, P., 2014. Tropical conversion and agroforestry intensification. Proc. Natl. Acad. Sci. 12, 4973–4978.
forests in the Anthropocene. Annu. Rev. Env. Resour. 39, 125–159. Stotz, D.F., Fitzpatrick, J.W., Parker III, T.A., Moskovits, D.K., 1996. Neotropical Birds:
Ecology and Conservation. University of Chicago Press, Chicago, p. 482.

8
J.P. Cabral et al. Forest Ecology and Management 481 (2021) 118704

Stouffer, P.C., Bierregaard, R.O., 1995. Use of Amazonian forest fragments by understory Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S, Fourth edition.
insectivorous birds. Ecology 76, 2429–2445. Springer, New York, p. 516.
Thomas, W.W., Carvalho, A.M.V., Amorim, A.M., Garrison, J., Arbelaez, A.L., 1998. Plant Waltert, M., Bobo, K.S., Sainge, N.M., Fermon, H., Mühlenberg, M., 2005. From forest to
endemism in two forests in southern Bahia. Brazil. Biodiver. Conserv. 7, 311–322. farmland: habitat effects on Afrotropical forest bird diversity. Ecol. Appl. 15,
Tscharntke, T., Sekercioglu, C.H., Dietsch, T.V., Sodhi, N.S., Hoehn, P., Tylianakis, J.M., 1351–1366.
2008. Landscape constraints on functional diversity of birds and insects in tropical Watling, J.I., Arroyo-Rodríguez, V., Pfeifer, M., Baetenm, L., Banks-Leite, C., Cisneros, L.
agroecosystems. Ecology 89, 944–951. M., Fang, R., Hamel-Leigue, A.C., Lachat, T., Leal, I.R., Lens, L., Possingham, H.P.,
Tscharntke, T., Tylianakis, M., Rand, T.A., Didham, R.K., Fahrig, L., Batáry, P., Raheem, D.C., Ribeiro, D.B., Slade, E.M., Urbina-Cardona, J.N., Wood, E.M.,
Bengtsson, J., Clough, Y., Crist, T.O., Dormann, C.F., Ewers, R.M., Fründ, J., Holt, R. Fahrig, L., 2020. Support for the habitat amount hypothesis from a global synthesis
D., Holzschuh, A., Klein, A.M., Kleijn, D., Kremen, C., Landis, D.A., Laurance, W., of species density studies. Ecol. Lett. 23, 674–681.
Lindenmayer, D., Scherber, C., Sodhi, N., Steffan-Dewenter, I., Thies, C., van der Wilman, H., Belmaker, J., Simpson, J., De La Rosa, C., Rivadeneira, M., Jetz, W., 2014.
Putten, W.H., Westphal, C., 2012. Landscape moderation of biodiversity patterns and Elton Traits 1.0: Species-level foraging attributes oh the world’s birds and mammals.
processes – eight hypotheses. Biol. Rev. 87, 661–685. Ecology 95, 95–178.
Van Bael, S.A.V., Philpott, S.M., Greenberg, R., Bichier, P., Barber, N.A., Mooney, K.A., Wood, S. 2017. Generalized Additive Models: An Introduction with R, 2 edition.
Gruner, D.S., 2008. Birds as predators in tropical agroforestry systems. Ecology 89, Chapman and Hall/CRC. 496 p.
928–934. Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. Mixed Effects
Vandermeer, J.H., 2011. The Ecology of Agroecosystems. Jones and Bartlett Learning, Models and Extensions in Ecology with R. Springer, New York, p. 574.
Boston, p. 387.

You might also like