Content uploaded by Ugo Diniz
Author content
All content in this area was uploaded by Ugo Diniz on Nov 09, 2021
Content may be subject to copyright.
Biotropica. 2020;00:1–12.
|
1wileyonlinelibrary.com/journal/btp
Received: 18 March 2020
|
Revised: 2 Oc tober 2020
|
Accepted: 17 October 2020
DOI: 10.1111/btp.12902
ORIGINAL ARTICLE
Bats and hawkmoths form mixed modules with flowering
plants in a nocturnal interaction network
Joel A. Queiroz1 | Ugo M. Diniz2 | Diego P. Vázquez3,4 | Zelma M. Quirino5 |
Francisco A. R. Santos6 | Marco A. R. Mello7 | Isabel C. Machado8
© 2020 The Association for Tropical Biology and Conservation
1Departamento de Educação, Universidade
Federal da Paraíba, Mamanguape, Brasil
2Programa de Pós-Graduação em Ecologia,
Universidade de Brasília, Brasília, Brasil
3Instituto Argentino de Investigaciones de
las Zonas Áridas, Mendoza, Argentina
4Facultad de Ciencias Exactas y Naturales,
Universidad Nacional de Cuyo, Mendoza,
Argentina
5Departamento de Engenharia e Meio
Ambiente, Universidade Federal da Paraíba,
João Pessoa, Brasil
6Departamento de Ciências Biológicas,
Universidade Estadual de Feira de Santana,
Feira de Santana, Brasil
7Departamento de Ecologia, Universidade
de São Paulo, São Paulo, Brasil
8Departamento de Botânica, Universidade
Federal de Pernambuco, Recife, Brasil
Correspondence
Isabel C. Machado, Departamento de
Botânica, Universidade Federal de
Pernambuco, Recife, Brasil.
Email: imachado@ufpe.br
Funding information
Pernambuco Research Foundation (FACEPE),
Grant/Award Number: APQ-1096-
2.03/08; Brazilian Council for Scientific
and Technological Development (CNPq),
Grant/Award Number: 459485/2014-
8, 302700/2016-1, 304498/2019-0,
18529/12-7 and 311021/2014-0; Alexander
von Humboldt Foundation (AvH), Grant/
Award Number: 3.4-8151/15037 and 3.2-
BRA/1134644; Dean of Research of the
University of São Paulo (PRP-USP), Grant/
Award Number: 18.1.660.41.7; São Paulo
Research Foundation (FAPESP), Grant/
Award Number: 2018/20695-7; Brazilian
Coordination for the Improvement of Higher
Education Personnel (C APES), Grant/Award
Number: 001
Associate Edito r: Tomás A. Carlo
Handling Editor: Nico Bluthgen
Abstract
Based on the conceptual framework of pollination syndromes, pollination networks
should be composed of well-delimited subgroups formed by plants that diverge in flo-
ral phenotypes and are visited by taxonomically different pollinators. Nevertheless,
floral traits are not always accurate in predicting floral visitors. For instance, flowers
adapted to bat-pollination are larger and wider, enabling the exploitation by other
nocturnal animals, such as hawkmoths. Thus, should an interaction network com-
prising bats and hawkmoths, the most important nocturnal pollinators in the tropics,
be formed of mixed-taxon modules due to cross-syndrome interactions? Here, we
analyzed such a network to test whether resource plants are shared between the two
taxa, and how modules differ in terms of species morphologies. We sampled interac-
tions through pollen grains collected from floral visitors in a Caatinga dry forest in
northeastern Brazil. The network was modular yet interwoven by cross-syndrome
interactions. Hawkmoths showed no restriction to visit the wider chiropterophilous
flowers. Furthermore, bats represented a subset of a hawkmoth-dominated network,
as they were restricted to chiropterophilous flowers due to constraints in accessing
narrower sphingophilous flowers. As such, the bat-dominated module encompassed
relatively wider flowers, but hawkmoths, especially long-tongued ones, were unre-
stricted by floral width or length. Thus, pollination of flowers with open architectures
must be investigated with caution, as they are accessible to a wide array of floral
visitors, which may result in mixed-pollination systems. Future research should con-
tinue to integrate different syndromes and pollinator groups in order to reach a bet-
ter understanding of how pollination-related functions emerge from community-level
interactions.
Abstract in Portuguese is available with online material.
KEY WORDS
Caatinga, chiropterophily, mixed-pollination, modularity, pollination syndromes, specialization,
sphingophily
2
|
QUEIROZ Et al .
1 | INTRODUCTION
Plants with specialized phenotypes and highly derived floral traits
tend to make ecologically specialized interactions and rely on a
few effective pollinators (e.g., Johnson & Steiner, 1997; Manning &
Goldblatt, 1997). However, adaptations to multiple guilds of pollina-
tors that exert divergent selective pressures are far more common
in nature than previously thought (Fenster et al., 2004). Thus, plants
often show mixed-pollination systems, and many are ecological gen-
eralists (Armbruster, 2017).
Those mixed systems challenge the concept of pollination syn-
dromes (sensu Faegri & Pijl, 1979), that is, the conceptual framework
that predicts the most effective pollinator group of a plant based
on its set of floral traits. Nevertheless, the concept of pollination
syndromes is often seen as an effective framework, especially when
considering that syndromes are not restricted to predicting only the
most effective pollinator groups, but also secondary ones (Rosas-
Guerrero et al., 2014).
Although this concept has met criticism regarding its useful-
ness to understand some systems (Ollerton et al., 2009), noctur-
nal pollination syndromes seem to be especially good predictors of
real-world interactions (Muchhala & Jarrin, 2002). Flower-visiting
bats (Chiroptera: Phyllostomidae) and hawkmoths (Lepidoptera:
Sphingidae) are the two largest guilds of specialized nocturnal pol-
len vectors in the Neotropics (Borges et al., 2016). In addition, the
pollination syndromes associated with them (chiropterophily and
sphingophily, respectively) differ markedly in floral phenotype (Tripp
& Manos, 2008). Sphingophilous flowers tend to be more restrictive
than chiropterophilous flowers, and extreme cases of morphologi-
cal specialization are commonly found among the former (Martins &
Johnson, 2007; Nilsson, 1988). Long and narrow floral tubes match-
ing the length of the hawkmoths’ feeding apparatus are a convergent
feature of sphingophilous species and greatly constrain the visita-
tion by other pollinator guilds (Johnson et al., 2017). However, many
species with open floral morphologies and exposed nectar are also
pollinated by hawkmoths, which highlights that flowers pollinated
by these insects do not always conform to a predictable architecture
(Amorim et al., 2013; Koptur, 1983).
Open floral morphologies are a diagnostic trait of chiropteroph-
ilous plants, among which cup-, brush-, or bell-shaped flowers are
the most common (Fleming et al., 2009). Although such floral types
that confo rm to chi roptero phily allow th e interact ion wit h large r pol-
linators such as bats, they also enable flowers to be exploited and
pollinated by animals of other guilds, such as hummingbirds, bees,
and hawkmoths (Amorim et al., 2013; Queiroz et al., 2016; Rocha
et al., 2020). Exclusive dependence of flowers on bats, resulting
in an exclusion of other pollinators through morphological restric-
tion, does exist within the chiropterophilous syndrome but is rare
(Muchhala, 2006; Muchhala & Thomson, 2009).
Due to the lack of morphological constraints of bat-pollinated
plants, mainly those with open flowers, bats and hawkmoths likely
share some food-plants. Ecologists have recently started to quantify
community-wide patterns of interaction in bat–plant (Sritongchuay
et al., 2019; Stewart & Dudash, 2017) and hawkmoth–plant (Johnson
et al., 2017; Sazatornil et al., 2016) pollination networks, but we still
do not know the structure of the networks formed of both pollina-
tor guilds together. To assess specialization in those mixed systems,
we can use a network approach to gain insights on the structural
characteristics of community-wide mutualistic interactions, such
as modularity (Olesen et al., 2007). Modular networks are formed
by subgroups of species that interact more among themselves than
with the rest of the community (Dupont & Olesen, 2009). In pollina-
tion networks, modularity can reveal patterns of niche specialization
between interacting species (Amorim, 2020; Maruyama et al., 2014;
Phillips et al., 2020).
Therefore, in the present study, we aimed at unveiling the
structure of an interaction network formed by plants, bats, and
hawkmoths. Because most sphingophilous species present a more
restrictive floral morphology than chiropterophilous species, we ex-
pected hawkmoths to exploit plants of both pollination syndromes,
while bats should be constrained to chiropterophilous flowers. If this
is true, first, the network should have a modular structure, with more
modules including only sphingophilous plants and hawkmoths, and
fewer mixed modules including chiropterophilous plants, bats, and
hawkmoths. Second, if sphingophilous flowers restrict bat access
through morphology, the gradient of morphological restriction im-
posed by flowers should be associated with module configuration.
Thus, modules should differ in terms of floral tube characteristics
and visitor tongue length, with hawkmoth-dominated modules con-
taining longer and narrower flowers, and longer-tongued hawkmoths
placed alongside longer-tubed flowers. Finally, sphingophilous plants
should show higher species-level specialization (fewer mutualistic
partners) than chiropterophilous plants, while the latter should be
more generalist and stronger network connectors.
2 | MATERIALS AND METHODS
2.1 | Study area
Expeditions for the sampling of hawkmoths and bats were carried
out in 2011 and 2012 in the Almas Farm (7°28′45″ S, 36°54′18″
W), a private protected area located in the Cariris Velhos region,
Paraíba state, northeastern Brazil. This is an area of hiperxerophil-
ous Caatinga, with an average annual rainfall of 674.22 mm concen-
trated in three to five months a year, usually from January to May,
which results in a long dry period (Brazilian National Institute of
Meteorology [INMET], 2020).
Almas Farm has 3,505 ha of bush-dominated, deciduous veg-
etation, with a predominance of succulent components (Andrade-
Lima, 1981; Rodal & Sampaio, 2002). As already reported for other
Caatinga areas (Machado & Lopes, 2004), the Almas Farm has a rel-
atively high frequency of moth-pollinated (13%) and bat-pollinated
species (11%) (Quirino & Machado, 2014).
|
3
QUEIROZ Et al .
2.2 | Plant–pollinator interactions
Interactions between plants and nocturnal floral visitors were as-
sessed through pollen types found on the body of floral visitors. We
sampled pollen loads on the bodies of bats and hawkmoths captured
in the field, thus following a zoocentric approach (Jordano, 2016a;
Jordano et al., 2009). Because we considered only these two most
relevant pollinator groups in the region, other functional groups
such as noctuid moths, nocturnal bees, and non-volant mammals
were not sampled. Captures were carried out monthly from January
2011 to December 2012, from three to five nights per month, be-
tween 1800 hr and 0400 hr each night. Bats were captured with a
12 × 3-m mist net in the first 10 nights (1,000 mist-net hr) and with
four 7 × 2.5-m mist nets in all subsequent nights (2,480 mist-net hr)
(all nets with a 16 mm mesh size, 75/2 denier, Ecotone®). Nets were
set at ground level.
Because the study site is composed mostly of open vegetation,
the mist-nets were set atop rock outcrops, which contained denser
patches of flowering plants, in order to increase capture rates. The
patches often included both chiropterophilous and sphingophi-
lous plants flowering simultaneously. All captured individuals of
both specialized nectar-feeding bats (Phyllostomidae: subfamilies
Glossophaginae and Lonchophyllinae) and opportunistic flower-vis-
iting species were sampled for pollen. Pollen grains were collected
from the external body of bats (head, abdomen, back, wings, and
uropatagium) with cubes of stained glycerinated jelly, which were
later mounted on microscope slides for identification through ex-
ternal morphology (Voigt et al., 2009). To reduce the risk of sample
contamination, jelly cubes were always kept in isolated vials, and
handling materials were sterilized between pollen samplings from
different floral visitors. Following the sampling procedure, bats were
marked with wing punches and then released. Recaptured individu-
als were not included in the sampling.
Hawkmoths were captured using light traps, which were also
mounted atop rock outcrops during two new-moon nights each
month as a means to increase trap attractiveness, adding up to 48
capture nights. One light trap was set up per night also from 18:00 hr
to 04:00 hr, thus resulting in 480 hr of passive trapping. Hawkmoth
and bat captures were carried out on consecutive nights and never
simultaneously. After capture, each hawkmoth specimen was sac-
rificed and deposited individually in paper envelopes. Pollen grains
were later collected from the external body and the stretched pro-
boscis of hawkmoths using the same method as described above for
bats.
All species of floral visitors and pollen types were identified to
the lowest possible taxonomic level. Pollen types found on floral
visitors were compared to a personal reference pollen collection
composed of plants from the Almas Farm and surrounding area,
as well as to the palynological collection at Feira de Santana State
University, Bahia, Brazil. Each pollen type was classified as either
“chiropterophilous” or “sphingophilous,” based on the syndrome
of the plant of origin. We used the classification of Quirino and
Machado (2014), which listed the syndromes found in the study
area. The authors classified both sphingophilous and phalenophilous
(pollination by noctuid moths) species as “phalenophilous,” following
the terminology used by Faegri and Pijl (1979), who included both
groups in the same guild. Because we did not include noctuid moths
in our sampling, these species found interacting with hawkmoths in
the present work were referred to as “sphingophilous.” Pollen loads
could also contain pollen from plant species with syndromes other
than our focal chiropterophilous or sphingophilous systems. These
pollen types (e.g., diurnal species with anthesis at dusk or dawn)
were grouped into the third category “other syndromes.” Less than
10 grain of a pollen type present in a given sample were considered
contamination, as well as pollen from anemophilous species (e.g.,
family Poaceae).
Because some night-blooming plant species from the study area
start producing nectar in the late afternoon or end nectar production
by early morning, individuals of diurnal hawkmoths may also be at-
tracted to them during early or late anthesis. Therefore, additionally
to captures by light traps, we also performed daytime observations
(17:00 hr to 18:30 hr) to nocturnal flowers whose nectar dynamics
spans beyond nighttime in order to complement the network with
these potential diurnal interactions. Observations were made to the
sphingophilous Guettarda angelica (Rubiaceae), 9 hr; Aspidosperma
pyrifolium (Apocynaceae), 9 hr; Amburana cearensis (Fabaceae), 9 hr;
and to the chiropterophilous Ipomoea vespertilia (Convolvulaceae),
12 hr; Ceiba glaziovii (Malvaceae), 12 hr; Encholirium spectabile
(Bromeliaceae), 12 hr; Helicteres baruensis (Malvaceae), 9 hr. The fol-
lowing species were also observed early in the morning (05:30 hr
to 07:00 hr): the chiropterophilous Pseudobombax marginatum
(Malvaceae), 6 hr; and Pilosocereus gounellei (Cactaceae), 12 hr; and
the sphingophilous Cereus jamacaru (Cactaceae), 12 hr.
Because our sampling involves very distinct groups, we analyzed
sampling separately for bats and hawkmoths following Chacoff
et al. (2012). We estimated rarefaction curves of pollen types found
on bats and hawkmoths (nocturnal only) using an individual-based
approach (Colwell et al., 2012), that is, by quantifying pollen types
accumulated according to the total number of interactions recorded
from captures visitor individuals. We used the nonparametric esti-
mator of asymptotic species richness Chao1, which takes into ac-
count interaction abundances (Chao et al., 2009), to calculate the
expected richness of pollen types for both curves. Sampling com-
pleteness was calculated as the percentage of the observed richness
of pollen types in each curve in relation to the estimated asymptoti-
cal richness (Chacoff et al., 2012).
2.3 | Network structure
Data on the pollen types found on floral visitors were used to build a
weighted incidence matrix of interactions (A × B), in which rows cor-
respond to flor al visitor s (i) and colu mns , to pl ant s (j). Each matrix cell
aij contained values of interaction frequencies, that is, the number
of times a floral visitor species i was found carrying the pollen of a
plant species j.
4
|
QUEIROZ Et al .
We did not consider the absolute number of pollen grains found
on each individual floral visitor, but rather the presence or absence
of pollen. Interaction frequencies, therefore, represented the num-
ber of individuals of floral visitor species i that carried pollen of plant
species j. Weighted matrices are considered more informative than
their binary counterparts for assessing local realized niches (Fründ
et al., 2016), as links in the former are better proxies for recipro-
cal effects among species (Vázquez et al., 2005). Moreover, several
weighted network metrics are less biased by sampling incomplete-
ness than their binary versions (Vizentin-Bugoni et al., 2016).
We described the structure of the studied network using three
network-level metrics. Complementary specialization (H2’) is a
measure of niche divergence between species and varies between
0 and 1, where higher values of H2’ indicate higher specialization
(Blüthgen, 2010). This measure is considered a good proxy for eco-
logical specialization at the network level. Nestedness, assessed
through the WNODF metric (Almeida-Neto & Ulrich, 2011), de-
scribes to which extent the interactions of species with fewer mu-
tualistic partners form a subset of the interactions of species with
more mutualistic partners. WNODF ranges from 0 (non-nested net-
work) to 1 (perfectly nested network).
Weighted modularity (Qw) assesses the extent to which species
form subgroups with higher internal than external interaction den-
sity. This metric is a good proxy for ecological concepts such as guild
and functional group. We used the Beckett weighted modularity
maximizing algorithm DIRTLPAwb+ (Beckett, 2016), which com-
putes faster, has more stable results, and is less sensitive to initial
node labeling than previous algorithms used in the ecological liter-
ature (such as Guimerà, Barber, QuanBiMo, LPAb+, and LPAwb+).
This algorithm assigns unique labels (modules) to each node from the
smallest of the two species sets (higher or lower level) and then runs
a sequence of steps of label propagation followed by module joining
until the greatest value of modularity is achieved. The procedure is
run multiple times with different label initializations. The values of
weighted modularity (Qw) range from 0 (no modular structure) to 1
(high density of heavier links within the modules). We made 100 it-
erations of the algorithm and reported the modular structure that
resulted in the highest value of Qw.
Finally, we also tested for a compound topology (sensu
Lewinsohn et al., 2006) in the studied network. A compound
network has a modular structure, but its modules show a differ-
ent kind of internal structure, such as nestedness (see Pinheiro
et al., 2019). We used a set of customized functions written in R
to run this analysis (see Felix et al., 2017; Pinheiro et al., 2019, and
the supplement). With these functions, we calculated nestedness
in the entire net work, bet ween it s mod ules, and within it s mod ules.
For this analysis, we used the WNODA metric, which is derived
from WNODF, but focuses on decreasing marginal totals instead
of decreasing fill, and allows analyzing completely filled matrices
(Pinheiro et al., 2019). A compound network is expected to show
higher nestedness within its modules than between its modules
and in the entire network.
The significance of the network metrics H2
’, WNODF, WNODA,
and QW was estimated through Monte Carlo procedures based on
comparisons with randomized distributions generated with null
models. We generated 1,000 randomized matrices based on the
original weighted matrix using the algorithm proposed by Vázquez
et al. (2007) for pollination networks, which assigns interac tions ran-
domly among pairs of species with the constraints that the matrix
size and connectance remain the same as in the original matrix. The
algorithm has an intermediate level of restriction as the randomized
marginal totals vary from those of the original network (Dormann
et al., 2019). For estimating the significance of WNODA at differ-
ent network scales, we used the restricted null model (Pinheiro
et al., 2019), which is an adaptation of the Vázquez null model for
using with WNODA.
Significance (p) was estimated as the number of times the null
model resulted in a network with a score equal or higher than the
score measured in the original matrix, divided by the total number
of randomizations (1,000). We report p-values exactly as calculated,
and not as approximations. Network structure was considered to
significantly deviate from the null model when p≤0.05.
2.4 | Species roles
To assess the relative importance of each node for the structure of
the network, we calculated a set of species-level metrics. The cen-
trality of a species was measured first as its normalized degree (nk),
that is, the proportion of partners that a given species interacts
with out of the total number of potential partners available in the
network (Freeman, 1979). A species connected to a larger propor-
tion of partners is assumed to be more influential in the structure
and dynamics of its network (González et al., 2010). We also calcu-
lated betweenness centrality (BC), that is, the proportion of short-
est paths that pass through a node (Freeman, 1977), which varies
from 0 to 1. A species positioned between several pairs is assumed
to contribute more to connecting different regions of the network
(Mello et al., 2015). Finally, we quantified node specialization using
Blüthgen's d’, which measures the specialization of a node to a set of
other nodes and varies from 0 to 1 (Blüthgen, 2010). A more special-
ized node (d’ → 1) is assumed to represent a species that makes a set
of interactions different from those made by other species of the
same network (Mello et al., 2019).
These species-level metrics were compared among syndrome
categories, that is, chiropterophilous, sphingophilous, and oth-
er-syndrome plants by fitting generalized linear models (GLMs). We
fitted a separate GLM for each metric, employing metric values as
dependent variables and plant syndromes as the explanatory vari-
able. Models were fitted with a quasibinomial error distribution and
a logit link funct ion and were validated by comp aring them with null
models with no explanatory variables. Significance was assessed by
chi-squared tests. The significance of pairwise differences of met-
rics among syndromes was assessed through t tests after repeating
|
5
QUEIROZ Et al .
the GLMs once with each syndrome category as the baseline (inter-
cept) for comparison. We did not compare groups of floral visitors,
because there were many more hawkmoth species than bat species
in the network, which would lead to an unbalanced design.
2.5 | Drivers of module partitioning
Given a modular structure, we assessed whether the species’ mor-
phologies would differ according to the modules assigned by the
algorithm, and thus if visitor exclusion by morphological constraint
was a relevant process in shaping the network's structure. We
gathered data from the literature on the morphometry of struc-
tures related to floral restriction and to the access of pollinators
to flowers.
For pollinators, we collected data on the length of feeding ap-
paratus: proboscis length for hawkmoths (Johnson et al., 2017)
and operational tongue length—palate length plus tongue ex-
tension beyond mouth—for bats (Winter & Helversen, 2003).
Non-specialized nectarivorous bats, whose tongues do not ex-
tend significantly beyond the mouth, had only their palate length
considered.
For plants, we collected data on floral tube length and diameter
(see Table S1 for references and comments on the structures mea-
sured for each species). Whenever we could not find the data for a
species in the literature, we extracted the measurements ourselves
by compiling scaled photographic records of exsiccatae of the spe-
cies uploaded on the Species Link data base (http://www.splink.
org.br/) and using linear morphometric analysis in the ImageJ 1.8.0
software (Abràmoff et al., 2004). In these cases, 30 flowers were
used for measurement (see Table S1 for the examined material). Four
species from the “other syndromes” category, whose pollen types
that could not be identified to the lowest possible level, could not be
included in this analysis.
We ran a GLM with a Gaussian error distribution for each morpho-
logical variable, in which the measurements (tube length, tube diame-
ter, or tongue length) were set as the dependent variable, and module
assignment as the explanatory variable. The significance of the models
and pairwise differences among modules were assessed through the
same procedure described above at the Species roles topic.
2.6 | Software
All analyses were run in R 4.0.2 (R Development Core Team, 2020).
The computation of network metrics and the null model analysis
were made using the package bipar tite (Dormann et al., 2019), and
the sampling completeness analysis was performed using the pack-
age vegan (Oksanen et al., 2013). The compound topology analysis
was made using customized R functions (Felix et al., 2017; Mello
et al., 2019; Pinheiro et al., 2019), which we made available in the
supplement. The software Gephi 0.9.2 (Bastian et al., 2009) was
used for graph drawing.
3 | RESULTS
3.1 | Plant–pollinator interactions
We analyzed the pollen samples of 117 bats of four species and 246
hawkmoths of 15 species and recorded 766 interactions with 24
plant species (see Tables S1 and S2 for a complete list of species).
Out of the 15 hawkmoth species, Aellopos sp. (four individuals) was
the only diurnal recorded visiting plants during dusk.
We recorded five plant species classified as sphingophilous
and nine as chiropterophilous, based on diagnostic floral traits
(see Figure 1 for examples). Eight additional pollen types that did
not belong to either chiropterophilous or sphingophilous species
were also found in at least one pollen sample and were included
in the “other syndromes” category (Table S2). Two pollen types
remained unidentified. Two species from the pollen pool were not
studied by Quirino and Machado (2014): Hippeastrum aff. elegans
(Amaryllidaceae) and Cordiera aff. rigida (Rubiaceae). The latter has
small and inconspicuous flowers and is reported to be visited by
diurnal bees (Lopes et al., 2016) and was thus placed in the “other
syndromes” category. The former does not have a formal register
of its reproductive biology. Because of its robust, white flowers,
which were visited mostly by bats (Figure 2), it was classified as
“chiropterophilous.”
None of the pollen type rarefaction curves reached the esti-
mated asymptotic species richness, although the observed richness
of pollen types all fell into the standard error intervals of the es-
timated richness (Figure S1). Bat and hawkmoth samplings yielded,
respectively, 92.9% and 99.1% of the expected asymptotic pollen
richness.
The term species, when associated with plants, will be used
henceforth to refer to each distinc t pollen morphotype, and not nec-
essarily to the lowest possible taxonomic level.
3.2 | Network structure
The network presented low to intermediate specialization (H’2 = 0.34,
Z = 10.36, p = 0), nestedness (WNDOF = 0.34, Z = −3.01, p = 1.0),
and modularity (Qw = 0.37, Z = 15.29, p = 0). In the compound to-
pology analysis, nestedness within the modules was two times
higher (WNODA = 0.60, p = 0.48) than between the modules
(WNODA = 0.30, p = 0.46) and in the entire network (WNODA = 0.38,
p = 0.47) (Figure S2). Therefore, the network shows specialization in
the interactions, although not high. In addition, this specialization is
organized mainly in subgroups. Finally, most of the network's nested-
ness comes from within its modules.
Four modules were identified in the network (Figure 2). All bat spe-
cies belonged to a single module, along with most plant species deemed
as chiropterophilous. The only exceptions were Bauhinia cheilantha and
Pilosocereus gounellei, which were visited to a similar extent by both bats
and hawkmoths. These two plant species were assigned to one of the
other three modules, all of which encompassed hawkmoths only. Two
6
|
QUEIROZ Et al .
hawkmoth species, Protambulyx strigilis and Eumorpha obscura, were also
placed in the module dominated by bats and chiropterophilous species.
3.3 | Species-level metrics
Specialization varied significantly among plant species with different
pollination syndromes (χ2 = 0.86, df = 19, p < 0.0001). Sphingophilous
plant species (d’ = 0.44 ± 0.07) were more specialized in their floral
visitors than chiropterophilous (d’ = 0.24 ± 0.09) (estimate = 0.93,
t = 3.86, p < 0.005) and other-syndrome species (d’ = 0.21 ± 0.09)
(estimate = 1.08, t = 4.25, p < 0.0005). Nevertheless, the latter two
did not differ from one another (p = 0.56; Figure 3a).
Plant species with different syndromes also differed signifi-
cantly from one another in terms of normalized degree (χ2 = 1.51,
df = 19, p < 0.001). Nevertheless, this difference was detected only
between other-syndrome species (nk = 0.14 ± 0.09) and chirop-
terophilous species (nk = 0.39 ± 0.21; estimate = 1 .37, t = 3.26,
p < 0.01). Sphingophilous species (nk = 0.29 ± 0.09) did not differ
from the other groups (p > 0.05 for both; Figure 3b).
Finally, pollination syndromes did not explain the performance of
a plant species as a network connector (χ2 = 0 .957, df = 19, p = 0.76;
Figure 3c). However, the chiropterophilous plants Pilosocereus
gounellei and Bauhinia cheilantha and the diurnal plant Croton sp.
stood out for reaching the highest betweenness centrality values in
the network (BC = 0.11).
3.4 | Module assignment and species morphologies
Two out of the three morphological variables assessed explained
part of the module assignment detected for plant and pollinator
FIGURE 1 An overview of some species and interactions recorded in a Caatinga dry forest in northeastern Brazil. Chiropterophilous
plants: Encholirium spectabile receiving visits from the specialized nectar-feeding bat Glossophaga soricina (a) and the unspecialized
Phyllostomus discolor (b); Pseudobombax marginatum (c) and Pilosocereus gounellei (d) interacting with the specialized Lonchophylla mordax;
Ipomoea vespertilia receiving a visit from G. soricina (e) and being approached by the hawkmoth Agrius cingulata (f); (g) frontal view of the bell-
shaped flower of Pilosocereus chrysostele. Sphingophilous plants: (h) Tocoyena formosa visited by Manduca rustica; (i) lateral view of the long,
tubular flower of Cereus jamacaru
|
7
QUEIROZ Et al .
species in the network. The length of the feeding apparatuses
of pollinators varied significantly among modules (χ2 = 11,162.0,
df = 3, p < 0.0001), with one hawkmoth module being the only
containing visitors with longer apparatuses than the other mod-
ules (p < 0.05 for all comparisons; Figure 4). It encompassed the
hawkmoths with the longest proboscis in the network: Agrius cin-
gulata, Manduca rustica, M. sexta, and Cocytius antaeus (>95 mm
length).
Fl oral tu be di amet er also di f fere d among mod ules ( χ2 = 530.96 ,
df = 3, p < 0.005), but only between the bat-dominated module,
which contained chiropterophilous species only, and a hawk-
moth-dominated module that contained small sphingophilous
flowers (i.e., Guettarda angelica, Aspidosperma pyrifolium) and in-
conspicuous flowers from other syndromes (estimate = 12.32,
t = 1. 87, p < 0.01). There were no other significant pairwise dif fer-
ences in floral diameter (p > 0.05 for all). Finally, floral tube length
did not differ among modules (χ2 = 17,772.0, df = 3, p = 0.06),
yet all long-tubed flowers were placed alongside long-proboscis
hawkmoths (Figure 4).
4 | DISCUSSION
In the present study, we show that pollination syndromes predict
only part of the structure of a nocturnal interaction network formed
by bats and hawkmoths. On the one hand, the morphologically re-
strict sphingophilous plants excluded bats and formed modules that
contained only hawkmoths. On the other hand, hawkmoths had no
constraint in visiting most chiropterophilous plants, which showed,
on average, lower interaction specialization toward the visitors in
the network than the sphingophilous plants.
Thus, although bats interacted strongly with chiropterophilous
plants, the only module that contained bats also included hawk-
moths, making bat-plant interactions a subset in a network domi-
nated by hawkmoths. This may be caused by a marked difference in
species richness between those taxa. Hawkmoths presented almost
four times more species and over twice captured individuals than
bats. Although this difference could be interpreted as a consequence
of the higher abundance of hawkmoths in nature compared to bats,
the different methods employed to sample visitors (passive for bats
FIGURE 2 The nocturnal plant–pollinator interaction network in a Tropical Dry Forest of northeastern Brazil. Pollinator nodes are
drawn in white, while plant nodes are colored according to pollination syndromes. Link width is proportional to interaction frequency. Gray
polygons around groups of nodes represent the interaction modules identified with the Beckett modularity detection algorithm. Pollinators:
1. Callionima grisescens; 2. Erinnyis ello; 3. Lonchophylla mordax; 4. Glossophaga soricina; 5. Isognathus allamandae; 6. Agrius cingulata; 7.
Eumorpha vitis; 8. Eumorpha analis; 9. Manduca rustica; 10. Phyllostomus discolor; 11. Artibeus planirostris; 12. Manduca sexta; 13. Xylophanes
tersa; 14. Protambulyx strigilis; 15. Erinnyis obscura; 16. Aellopos sp.; 17. Cocytius antaeus; 18. Callionima parse; 19. Pseudosphinx tetrio. Plants:
20. Pilosocereus gounellei; 21. Bauhinia cheilantha; 22. Ceiba glaziovii; 23. Encholirium spectabile; 24. Ipomoea vespertilia; 25. Guettarda angelica;
26. Tocoyena formosa; 27. Amburana cearensis; 28. Helicteres baruensis; 29. Croton sp.; 30. Cereus jamacaru; 31. Pseudobombax marginatum; 32.
Indet.1; 33. Hippeastrum aff. elegans; 34. Piptadenia stipulacea; 35. Pilosocereus chrysostele; 36. Aspidosperma pyrifolium; 37. Combretum sp.;
38. Anadenanthera colubrina; 39. Cordiera aff. rigida; 40. Allophylus sp.; 41. Poincianella sp.; 42. Schinopsis brasiliensis; 43. Indet.2
8
|
QUEIROZ Et al .
and active for hawkmoths) may be responsible for their contrast-
ing richness. Notwithstanding, even though bats would require a
larger additional sampling effort to reach the expected pollen type
richness compared to hawkmoths, as suggested by the rarefaction
curves, both groups reached a sufficiently high percentage of sam-
pling completeness by falling into the confidence intervals of the as-
ymptotical richness estimator.
This dominance of a visitor group has been observed in sin-
gle-guild nocturnal pollination networks. Long-tubed flowers ex-
clude short-proboscis hawkmoths while long-proboscis hawkmoths
can forage on a wider variety of partners (Moré et al., 2007). Similarly,
old-world opportunistic flower-visiting bats share only a subset
of the partners of specialized nectarivorous bats (Sritongchuay
et al., 2019). Our results also support that the modularity resultant
from a gradient of morphological restriction, which is recurrent in
plant–pollinator systems (Dalsgaard et al., 2009; Jordano, 2016b),
will often emerge also in systems comprising phylogenetically dis-
tant pollinators (e.g., Danieli-Silva et al., 2012).
In the studied network, floral width appears to pose a signifi-
cant yet weak gradient of morphological restriction, with hawk-
moth-dominated modules encompassing relatively narrower
flowers. This seems enough to block bat access, while the inverse
for chiropterophilous flowers allows interactions with both taxa.
Moreover, our results corroborate the wider potential niche of long-
tongued hawkmoths (Johnson et al., 2017), whose module presented
significantly longer feeding apparatuses coupled with a large varia-
tion in floral tube length, spanning from those of the long- and nar-
row-tubed C. jamacaru and T. fo rmo sa (specialists), to those of the
wide chiropterophilous B. cheilantha and P. gounellei (generalists).
Such variation was likely a key factor in homogenizing floral tube
lengths among modules.
Although the morphological gradients of flowers were not as
conspicuous as expected, the incidence of several unrealized inter-
actions derived from morphological matching was likely responsible
for the obser ved low sp ecial ization, modularit y, and nested ness, cre-
ating non-inclusive sets of interactions (Almeida-Neto et al., 2007).
Generalist bat species (i.e., higher degrees—L. mordax, G. soricina)
do not interact with any of the narrow and generalist sphingoph-
ilous plants due to their clear morphological restriction. Similarly,
the specialist, long-tubed sphingophilous plants (such as T. fo r-
mosa and C. jamacaru) interact more frequently with long-tongued
hawkmoths, a matching also observed by Sazatornil et al. (2016)
FIGURE 3 Distribution of species-level metrics for plants belonging to different pollination systems in the nocturnal pollination network.
sphin: sphingophilous species; chiro: chiropterophilous species; other: species with other pollination syndromes. nk: normalized degree; d’:
Blüthgen's specialization index; BC: betweenness centrality. Asterisks represent significant differences between groups (GLM)
FIGURE 4 Distribution of
morphological trait measurements for
plants (floral tube length and diameter)
and floral visitors (tongue length)
composing the nocturnal network,
classified by module assignment. Modules
are divided into a bat-dominated module
(bat) and three hawkmoth-dominated
modules (hawk1, hawk2, and hawk3)
|
9
QUEIROZ Et al .
and possibly a result of the energetical constraint of larger hawk-
moths that require the more abundant nectar of long-tubed flowers
(Johnson et al., 2017).
All those constraints and trait-matching processes might explain
the structure of the studied network. We found evidence that it has
a compound topology (sensu Pinheiro et al., 2019). In order words, it
has a modular structure, but its modules seem to be internally nested.
In fac t, th e network's nestedness comes mostly from within than be-
tween the modules. Compound topologies have also been observed
in other networks with high taxonomic diversity (Felix et al., 2017),
composed of different kinds of interactions (Mello et al., 2019), or
that contain mixed taxonomic groups (Genrich et al., 2017). As the
studied network does also harbor mixed taxa, its compound topol-
ogy might have emerged as a result of the constraints faced by tax-
onomically distant pollinators when using the resources offered by
different plant families and genera. Once a constraint is resolved and
the pollinator gets associated with a module, its interactions within
this module become more flexible, which allows nested associations
to emerge.
Some groups overcome those constraints though. For instance,
the ability of hawkmoths of exploiting plants belonging to dif ferent
pollination syndromes is consistent with the evidence that these
animals play the role of ecological generalists in their communities,
especially long-tongued ones, and reinforces their role as connec-
tors of nocturnal, and perhaps even diurnal, pollination systems
(Haber & Frankie, 1989; Nilsson et al., 1987). The exploitation by
hawkmoths of flowers with less restrictive shapes reached even
species belonging to other syndromes beyond chiropterophily, es-
pecially by small-tongued hawkmoths (e.g., Amorim, 2020). Much
like many chiropterophilous plants, species such as Anadenanthera
colubrina, and those from the genera Combretum and Croton exhibit
lush brush-like inflorescences. Based on studies with other species
of those genera, these are probably insect-pollinated but have an-
thesis that spans beyond daytime (e.g., Ekeke & Agbagwa, 2015;
Freitas et al., 2001), which could have enabled the visitation by
hawkmoths and also sporadically by bats. These species, however,
were mostly marginal in the network, receiving few visitors. This
was expecte d since diurnal visito rs were not the target of our work .
These species would likely have more central roles if sampled in a
different context.
Our results also suggest that hawkmoths are important compo-
nents of mixed-pollination in night-blooming plants. In plants with
brush-like flowers and generalized pollination systems, hawkmoths
can be, along with bats, among the most effective pollinators (Cruz-
Neto et al., 2015). Moreover, as revealed by our results, hawkmoths
potentially act as relevant secondary pollinators to species regarded
as chiropterophilous, or perhaps even as effective as bats (Rocha
et al., 2020). Here, we bring further evidence that chiropterophilous
plants with open floral architectures are especially prone to pres-
ent multiple groups of floral visitors and mixed systems (Queiroz
et al., 2016). Further works on chiropterophilous species should,
therefore, always consider coupling visitation data with pollinator
efficiency (e.g., Santiago-Hernández et al., 2019).
We reported certain species whose syndromes were not
good predictors of pollinator attraction and visitation. The spe-
cies Bauhinia cheilantha, classified as chiropterophilous (Quirino &
Machado, 2014), has a potential mixed-pollination system as it in-
teracted with bats and hawkmoths with similar intensities. The same
occurs for the hypergeneralist Pilosocereus gounellei, whose syn-
drome is a contentious subject as different authors have reported
different pollinator activity. Although deemed as chiropterophilous
and evidence of bat-pollination being found for it in one region
(Cordero-Schmidt et al., 2017), only visitation by hawkmoths has
been reported for it in another Caatinga area (Rocha et al., 2020). The
same dichotomy was observed for the chiropterophilous Lafoensia
pacari (Lythraceae), whose large cup-like flowers were visited by
hawkmoths only in a survey in Costa Rica (Haber & Frankie, 1989) and
exclusively by bats in the Brazilian Cerrado (Sazima & Sazima, 1975).
These contrasting visitor assemblages are certainly a result of local
pollinator assemblages, but we also suggest that these species do
not fall into simple syndrome categories. Pilosocereus gounellei, spe-
cifically, seems to act as a keystone species for nocturnal communi-
ties of floral visitors in the Caatinga. Its year-round flowering phase
(Quirino & Machado, 2014), and consequently the overlap with the
phenophases of both pollinator groups, may have also been a key
factor in homogenizing its interaction frequencies among guilds.
We also highlight a few unexpected interactions between plants
and hawkmoths found in the study site. The species Tocoyena for-
mosa, despite its very long and narrow tube (ca. 110 mm, Figure 1H),
interacted with three short-tongued hawkmoths (30–45 mm pro-
boscis). While the possibility of sample contamination cannot be
excluded, the short-tongued hawkmoths could in fact be attracted
by these flowers’ display or odor and try to feed on it, in which case
they could in fact act as effective pollinators even though they do
not consume nectar. Conversely, given if nectar accumulates inside
the floral tube of T. formo sa, short-tongued species could profit from
a portion of the resource. A similar case was reported for the short-
tongued hawkmoth Pachylia ficus (40 mm proboscis length), a legit-
imate pollinator of the long-spurred orchid Dendrophylax lindenii (at
least 120 mm spur length) in North America (Houlihan et al., 2019).
The short-tongued P. ficus possibly consumes a smaller quantity of
accumulated nectar in comparison with Cocytius antaeus, the long-
tongued pollinator of D. lindenii with a matching proboscis length.
Given the uncertainty surrounding these interactions with T. for-
mosa, we suggest in-depth investigations of these morphological
mismatches in future works.
Finally, we underline that we do not assume that hawkmoths
pollinated the chiropterophilous plants that they visited, as we did
not measure the efficiency of floral visitors. Our conclusions are
based on pollinator attraction and visitation to flowers, which can
but not necessarily result in effective pollination. Hawkmoths may
frequently visit large brush- or cup-shaped flowers, that is, typi-
cal chiropterophilous flowers, without touching their reproductive
parts or just contacting their anthers, thus not acting as effec-
tive pollinators of these plants (e.g., Domingos-Melo et al., 2019;
Machado et al., 2006). Visitation frequency was taken as a measure
10
|
QUEIROZ Et al .
of the intensity of a visitor's dependence on a given plant resource,
but it is not a good surrogate for assessing plant reliance on floral
visitors for pollination. Careful observations of pollen transfer must
be carried out before classifying floral visitors as pollinators, as pol-
lination networks tend to shift in structure after the exclusion of in-
efficient pollinators (Santiago-Hernández et al., 2019). Reliable data
on the species’ natural history are thus necessary to avoid making
biased conclusions about the structure and dynamics of mutualistic
networks.
ACKNOWLEDGMENTS
This research was funded by the Pernambuco Research Foundation
(FACEPE, APQ-1096-2.03/08) and the Brazilian Council for
Scientific and Technological Development (CNPq, Universal
459485/2014-8). The Brazilian Coordination for the Improvement
of Higher Education Personnel granted JAQ a sandwich scholarship
to study in Instituto Argentino de Investigaciones de Zonas Aridas
(IADIZA) (CAPES, 18529/12-7) and UMD a master's scholarship
(88882.347259/2019-01), Argentina. CNPq granted JAQ a Ph.D.
scholarship (18529/12-7) and ICM a research grant (311021/2014-
0). MARM was funded by the Alexander von Humboldt
Foundation (AvH, 3.4-8151/15037 and 3.2-BRA/1134644), CNPq
(302700/2016-1 and 304498/2019-0), Dean of Research of the
University of São Paulo (PRP-USP, 18.1.660.41.7), and São Paulo
Research Foundation (FAPESP, 2018/20695-7). UMD was granted
a graduate scholarship. This study was financed in part by the
Brazilian Coordination for the Improvement of Higher Education
Personnel (CAPES—Finance Code 001). José A. Duarte identified
th e hawkmo ths . Deoc lécio Q. Gu err a (in memoriam), Enrico Ber nar d,
and Juliana C. Correia identified the bats. Vanessa Nobrega helped
us in the field. Rober to Lim a provided us with logistic supp or t in the
field. The owners and managers of RPPN Fazenda Almas allowed us
to carry out the project on their farm. James Lucas helped us iden-
tify the Am aryllidacea e spe cies. The Lon g-Term Ec olo gic al Rese arc h
Program (PELD/CNPq, 52.0062./2006-0) provided us with logistic
support in the field. We are thankful to Prof. Felipe Amorim and to
an anonymous referee for their careful and enriching review of our
initial manuscript.
CONFLICT OF INTEREST
The authors declare no conflict of interest. The proper authoriza-
tion to capture, handle, and process animals was issued by the Chico
Mendes Biodiversity Conservation Institute—ICMBio (SISBIO no.
25582).
AUTHOR CONTRIBUTIONS
JAQ, ICM, and ZMQ conceptualized the study and designed the
methodology. ICM acquired the funding. JAQ carried out data cura-
tion and investigation under the supervision of ICM and ZMQ. FARS
identified the pollen types. UMD, JAQ, DPV, and MARM analyzed
the data. JAQ wrote the original draft of the manuscript, and UMD
wrote and reviewed the final version with the contributions of JAQ,
ICM, DPV, and MARM.
DATA AVA ILAB ILITY STATE MEN T
The processed data that support the findings of this study, as well
as the scripts used to analyze those data and produce the figures,
are available on GitHub (https://doi.org/10.5281/zenodo.4131207).
ORCID
Ugo M. Diniz https://orcid.org/0000-0003-3360-8314
Diego P. Vázquez https://orcid.org/0000-0002-3449-5748
Zelma M. Quirino https://orcid.org/0000-0003-0396-107X
Francisco A. R. Santos https://orcid.org/0000-0002-9246-3146
Marco A. R . Mello https://orcid.org/0000-0002-9098-9427
Isabel C. Machado https://orcid.org/0000-0001-5015-2393
REFERENCES
Abràmoff, M. D., Magalhães, P. J., & Ram, S. J. (2004). Image processing
with ImageJ. Biophotonics International, 11(7), 36–42.
Almeida-Neto, M., Guimarães, R. P. Jr, & Lewinsohn, M. T.
(2007). On nestedness analyses: Rethinking matrix tempera-
ture and anti-nestedness. Oikos, 116 (4), 716–722. https://doi.
org /10.1111/j.0 030-1299.2007.158 03 .x
Almeida-Neto, M., & Ulrich, W. (2011). A straightforward computational
approach for measuring nestedness using quantitative matrices.
Environmental Modelling & Software, 26(2), 173–178. https://doi.
org/10.1016/j.envso ft.2010.08.003
Amorim, F. W. (2020). Are the New World hummingbird-hawkmoths
functional equivalents of hummingbirds? Ecolog y, 101(12), e03161.
https://doi.org/10.10 02/ecy.3161
Amorim, F. W., Galetto, L., & Sazima, M. (2013). Beyond the pollination
syndrome: Nectar ecology and the role of diurnal and nocturnal pol-
linators in the reproductive success of Inga sessilis (Fabaceae). Plant
Biology, 15(2), 317–327.
Andrade-Lima, D. (1981). The Caatinga dominium. Revista Brasileira de
Botânica, 4, 149–163.
Armbruster, W. S. (2017). The specialization continuum in pollination
systems: Diversity of concepts and implications for ecology, evolu-
tion and conservation. Functional Ecology, 31(1), 88–100. https://doi.
org /10.1111/1365-2435.12783
Bastian, M., Heymann, S., & Jacomy, M. (2009, March). Gephi: An open
source software for exploring and manipulating networks. In Third
International AAAI Conference on Weblogs and Social Media.
Beckett, S. J. (2016). Improved community detection in weighted bipar-
tite networks. Royal Society Open Science, 3(1), 140536. https://doi.
org/10.1098/rsos.140536
Blüthgen, N. (2010). Why network analysis is often disconnected from
community ecology: A critique and an ecologist's guide. Basic
and Applied Ecology, 11 (3), 185–195. https://doi.org/10.1016/j.
baae.2010.01.001
Borges, R. M., Somanathan, H., & Kelber, A. (2016). Patterns and
processes in nocturnal and crepuscular pollination services.
The Quarterly Review of Biology, 91(4), 389–418. https://doi.
org/10.1086/689481
Brazilian National Institute of Meteorology (INMET). (2020). Historical
data from the Meteorological Database for Teaching and Research
[BDMEP]. Retrieved from http://www.inmet.gov.br/porta l/index.
php?r=bdmep/ bdmep
Chacoff, N. P., Vázquez, D. P., Lomáscolo, S. B., Stevani, E. L., Dorado, J.,
& Padrón, B. (2012). Evaluating sampling completeness in a desert
plant–pollinator network. Journal of Animal Ecology, 81(1), 190–200.
https://doi.or g/10.1111/ j.1365-2656.2011.018 83 .x
Chao, A., Colwell, R. K., Lin, C. W., & Gotelli, N. J. (20 09). Sufficient sam-
pling for asymptotic minimum species richness estimators. Ecology,
90(4), 1125–1133. https://doi.org/10.1890/07-2147.1
|
11
QUEIROZ Et al .
Colwell, R. K., Chao, A., Gotelli, N. J., Lin, S. Y., Mao, C. X., Chazdon, R.
L., & Longino, J. T. (2012). Models and estimators linking individu-
al-based and sample-based rarefaction, extrapolation and compari-
son of assemblages. Journal of Plant Ecology, 5(1), 3–21. https://doi.
org/10.1093/jpe/rtr044
Cordero-Schmidt, E., Barbier, E., Vargas-Mena, J. C., Oliveira, P. P.,
Santos, F. D. A. R., Medellín, R. A., Herrera, B. R., & Venticinque, E.
M. (2017). Natural history of the Caatinga endemic Vieira's flower
bat, Xeronycteris vieirai. Acta Chiropterologica, 19(2), 399–408.
Cruz-Neto, O., Machado, I. C., Galetto, L., & Lopes, A. V. (2015). The
influence of nectar production and floral visitors on the female re-
productive success of Inga (Fabaceae): A field experiment. Botanical
Journal of the Linnean Society, 177(2), 230–245.
Dalsgaard, B., González, A. M. M., Olesen, J. M., Ollerton, J., Timmermann,
A., Andersen, L. H., & Tossas, A. G. (2009). Plant–hummingbird inter-
actions in the West Indies: Floral specialisation gradients associated
with environment and hummingbird size. Oecologia, 159(4), 757–766.
https://doi.org/10.1007/s0044 2-008-1255-z
Danieli-Silva, A., de Souza, J. M. T., Donatti, A. J., Campos, R. P., Vicente-
Silva, J., Freitas, L., & Varassin, I. G. (2012). Do pollination syndromes
cause modularit y and predict interactions in a pollination network
in tropical high-altitude grasslands? Oikos, 121(1), 35–43. https://doi.
org /10.1111/j.1600 -0706.2011.190 89.x
Domingos-Melo, A., Milet-Pinheiro, P., Navarro, D. M. D. A. F., Lopes,
A. V., & Machado, I. C. (2019). It's raining fragrant nectar in the
Caatinga: Evidence of nectar olfactory signaling in bat-pollinated
flowers. Ecology, 101(3), e02914.
Dormann, C. F., Fründ, J., & Grüber, B. (2019). Package “bipartite”:
Visualizing bipartite net works and calculating some (ecological) indices.
Versio n, 2.13. Retrieved from https://cran.r-proje ct.org/web/packa
ges/bipar tite/bipar tite.pdf
Dupont, Y. L., & Olesen, J. M. (2009). Ecological modules and
roles of species in heathland plant–insect flower visitor net-
works. Journal of Animal Ecology, 78(2), 346–353. https://doi.
org /10.1111/j.1365-2656. 20 08.01501. x
Ekeke, C., & A gbagwa, I. O. (2015). Breeding and Pollination Biology of
Combretum constrictum (Benth) Laws. (Combretaceae). Tropical Plant
Biology, 8(3–4), 51–59. https://doi.org/10.1007/s1204 2-015-9151-2
Faegri, K., & van der Pijl, L. (1979). The principles of pollination ecology.
Pergamon Press.
Felix, G. M., Pinheiro, R. B. P., Poulin, R., Krasnov, B. R., & Mello, M. A.
R. (2017). The compound topology of a continent-wide interaction
network explained by an integrative hypothesis of specialization.
bioRxiv, 236687.
Fenster, C. B., Armbruster, W. S., Wilson, P., Dudash, M. R., & Thomson,
J. D. (2004). Pollination syndromes and floral specialization. Annual
Review of Ecology, Evolution, and Systematics, 35, 375–403. https://
doi.org/10.1146/annur ev.ecols ys.34.011802.132347
Fleming, T. H., Geiselman, C., & Kress, W. J. (2009). The evolution of bat
pollination: A phylogenetic perspective. Annals of Botany, 104(6),
1017–1043. ht tps://doi.org/10.1093/aob/mcp197
Freeman, L. C. (1977). A set of measures of centrality based on bet ween-
ness. Sociometry, 40(1), 35–41. https://doi.org/10.2307/3033543
Freeman, L. C. (1979). Centrality in social networks I: Conceptual clarifi-
cation. Social Networks, 1, 215–239.
Freitas, L., Bernardello, G., Galetto, L., & Paoli, A. A. (2001). Nectaries
and reproductive biology of Croton sarcopetalus (Euphorbiaceae).
Botanical Journal of the Linnean Society, 136 (3), 267–277. https://doi.
org /10.1111/j.1095-8 339.20 01.tb005 72.x
Fründ, J., McC an n, K. S., & Williams, N. M. (2016). Sampling bias is a chal-
lenge for quantifying specialization and network structure: Lessons
from a quantitative niche model. Oikos, 125(4), 502–513.
Genrich, C. M., Mello, M. A. R., Silveira, F. A. O., Bronstein, J. L., & Paglia,
A. P. (2017 ). Duality of interaction outcomes in a plant-f rugivo re mul-
tilayer network. Oikos, 126 , 361–368.
González, A. M. M., Dalsgaard, B., & Olesen, J. M. (2010). Centrality
measures and the impor tance of gene ra list specie s in pollination net-
works. Ecological Complexity, 7(1), 36–43.
Haber, W. A., & Frankie, G. W. (1989). A tropical hawkmoth community:
Costa Rican dr y forest Sphingidae. Biotropica, 155–172.
Houlihan, P. R., Stone, M., Clem, S. E., Owen, M., & Emmel, T. C . (2019).
pollination ecology of the ghost orchid (Dendrophylax lindenii): A first
description with new hypotheses for Darwin’s orchids. Scientific
Reports, 9(1), 1–10.
Johnson, S. D., Moré, M., Amorim, F. W., Haber, W. A., Frankie, G. W.,
Stanley, D. A., & Raguso, R. A. (2017). The long and the short of it: A
global analysis of hawkmoth pollination niches and interaction net-
works. Functional Ecology, 31(1), 101–115.
Johnson, S. D., & Steiner, K. E. (1997). Long-tongued fly pollination
and evolution of floral spur length in the Disa draconis complex
(Orchidaceae). Evolution, 51(1), 45–53.
Jordano, P. (2016a). Sampling networks of ecological interactions.
Functional Ecology, 30(12), 1883–1893.
Jordano, P. (2016b). Natural histor y matters: How biological constraints
shape diversified interactions in pollination networks. Journal of
Animal Ecology, 85(6), 1423–1426.
Jordano, P., Vázquez, D., & Bascompte, J. (2009). Redes complejas de in-
teracciones planta-animal. In R. Medel, M. Aizen, & R. Zamora (Eds.),
Ecología y evolución de las interacciones planta-animal: Conceptos y
aplicaciones (pp. 17–41). Editorial Universitaria.
Koptur, S. (1983). Flowering phenology and floral biology of Inga
(Fabaceae: Mimosoideae). Systematic Botany, 8(4), 354–368. https://
doi.org/10.2307/2418355
Lewinsohn, T. M., Inácio Prado, P., Jordano, P., Bascompte, J., & Olesen, J.
M. (2006). Structure in plant-animal interaction assemblages. Oikos,
113, 174–184. https://doi.org/10.1111/j.0030-1299.2006.14583.x
Lopes, C. G. R., Beirão, D. C. C., Pereira, L. A., & Alencar, L. C. (2016).
Levantamento da flora apícola em área de cerrado no município
de Floriano, estado do Piauí, Brasil. Revista Brasileira de Biociências,
14(2), 102–110.
Machado, I. C., & Lopes, A. V. (2004). Floral traits and pollination systems
in the Caatinga, a Brazilian tropical dry forest. Annals of Botany, 94(3),
365–376. https://doi.org/10.1093/aob/mch152
Machado, I. C., Lopes, A . V., Leite, A. V., & Neves, C. B. (2006).
Cleome spinosa (Capparaceae): Polygamodioecy and pol-
lination by bats in urban and Caatinga areas, northeast-
ern Brazil. Botanische Jahrbücher, 127(1), 69–82. https://doi.
org/10.1127/0006-8152/2006/0127-0069
Manning, J. C., & Goldblatt, P. (1997). The Moegistorhynchus longiros-
tris (Diptera: Nemestrinidae) pollination guild: Long-tubed flowers
and a specialized long-proboscid fly pollination system in southern
Africa. Plant Systematics and Evolution, 206(1–4), 51–69. ht tps ://do i.
org/10.1007/BF009 87941
Martins, D. J., & Johnson, S. D. (2007). Hawkmoth pollination of aeran-
goid orchids in Kenya, with spe cial reference to nectar suga r concen-
tration gradients in the floral spurs. American Journal of Botany, 94(4),
650–659. https://doi.org/10.3732/ajb.94.4.650
Maruyama, P. K., Vizentin-Bugoni, J., Oliveira, G. M., Oliveira, P. E., &
Dalsgaard, B. (2014). Morphological and spatio-temporal mismatches
shape a neotropical savanna plant-hummingbird network. Biotropica,
46(6), 740–747. https://doi.org/10.1111/btp.12170
Mello, M. A. R., Felix, G. M., Pinheiro, R. B. P., Muylaert, R. L., Geiselman,
C., Santana, S. E., Tschapka, M., Lotfi, N., Rodrigues, F. A., & Stevens,
R. D. (2019). Insights into the assembly rules of a continent-wide
multilayer network. Nature Ecology & Evolution, 3(11), 1525–1532.
https://doi.org/10.1038/s4155 9-019-1002-3
Mello, M. A. R., Rodrigues, F. A., Costa, L. D. F., Kissling, W. D., Şekercioğlu,
Ç. H., Marquitti, F. M. D., & Kalko, E. K. V. (2015). Keystone species
in seed dispersal networks are mainly determined by dietary special-
ization. Oikos, 124(8), 1031–1039. https://doi.org/10.1111/oik.01613
12
|
QUEIROZ Et al .
Moré, M., Sérsic, A. N., & Cocucci, A. A. (2007). Restriction of polli-
nator assemblage through flower length and width in three long-
tongued hawkmoth–pollinated species of Mandevilla (Apocynaceae,
Apocynoideae). Annals of the Missouri Botanical Garden, 94(2),
485–50 4.
Muchhala, N. (2006). Nectar bat stows huge tongue in its rib cage.
Nature, 444(7120), 701–702.
Muchhala, N., & Jarrin-V, P. (2002). Flower visitation by bats in Cloud
Forests of Western Ecuador1. Biotropica, 34(3), 387–395. https://doi.
org /10.1111/j.1744 -7429.2002.t b005 52. x
Muchhala, N., & Thomson, J. D. (20 09). Going to great lengths: Selection
for long corolla tubes in an extremely specialized bat–flower mutual-
ism. Proceedings of the Royal Society B: Biological Sciences, 276(1665),
2147–2152.
Nilsson, L . A. (1988). The evolution of flowers with deep corolla tubes.
Nature, 334(6178), 147–149.
Nilsson, L. A., Johnsson, L., Ralison, L., & Randrianjohany, E. (1987).
Angraecoid orchids and hawkmoths in central Madagascar:
Specialized pollination systems and generalist foragers. Biotropica,
310–318. https://doi.org/10.2307/2388628
Oksanen, J., Blanchet, F. G., Kindt, R., Legendre, P., Minchin, P. R., O'hara,
R. B., … Oksanen, M. J. (2013). Package ‘vegan’. Community Ecology
Package, 2(9), 1–295.
Olesen, J. M., Bascompte, J., Dupont, Y. L., & Jordano, P. (2007). The
smallest of all worlds: Pollination networks. Journal of Theoretical
Biology, 240 (2), 270–276. https://doi.org/10.1016/j.jtbi.2005.09.014
Ollerton, J., Alarcón, R., Waser, N. M., Price, M. V., Watts, S., Cranmer,
L., Hingston, A., Peter, C. I., & Rotenberry, J. (2009). A global test of
the pollination syndrome hypothesis. Annals of Botany, 103(9), 1471–
1480. https://doi.org/10.1093/aob/mcp031
Phillips, R. D., Peakall, R., van der Niet, T., & Johnson, S. D. (2020). Niche
perspectives on plant–pollinator interactions. Trends in Plant Science,
25(8), 779–793. https://doi.org/10.1016/j.tplan ts.2020.03.009
Pinheiro, R. B. P., Felix, G. M. F., Dormann, C. F., & Mello, M. A. R. (2019).
A new model explaining the origin of different topologies in interac-
tion networks. Ecology, 100(9), e02796.
Queiroz, J. A., Quirino, Z. G. M., Lopes, A. V., & Machado, I. C. (2016).
Vertebrate mixed pollination system in Encholirium spectabile: A bro-
meliad pollinated by bats, opossum and hummingbirds in a tropical
dry forest. Journal of Arid Environments, 125, 21– 3 0.
Quirino, Z. G. M., & Machado, I. C. (2014). Pollination syndromes in a
Caatinga plant community in northeastern Brazil: Seasonal avail-
ability of floral resources in different plant growth habits. Brazilian
Journal of Biology, 74(1), 62–71.
R Development Core Team. (2020). R: A language and environment for
statistical computing. The R Foundation for Statistical Computing.
Retrieved from https://www.r-proje ct.org/
Rocha, E. A., Domingos-Melo, A., Zappi, D. C., & Machado, I. C. (2020).
Reproductive biology of columnar cacti: Are bats the only protag-
onists in the pollination of Pilosocereus, a typical chiropterophilous
genus? Folia Geobotanica, 54, 239–256.
Rodal, M. J. N., & Sampaio, E. V. S. B. (2002). A vegetação do bioma
Caatinga. In E. V. S. B. Sampaio, A . M. Giulietti, J. Virgínio, & C. F. L.
G. Rojas (Eds.), Vegetação e flora da Caatinga (pp. 11–24). Associação
de Plantas do Nordeste.
Rosas-Guerrero, V., Aguilar, R., Martén-Rodríguez, S., Ashworth, L.,
Lopezaraiza-Mikel, M., Bastida, J. M., & Quesada, M. (2014). A quan-
titative review of pollination syndromes: Do floral traits predict ef-
fective pollinators? Ecology Letters, 17(3), 388–400.
Santiago-Hernández, M. H., Martén-Rodríguez, S., Lopezaraiza-Mikel,
M., Oyama, K., González-Rodríguez, A., & Quesada, M. (2019). The
role of pollination effectiveness on the attributes of interaction
networks: From floral visitation to plant fitness. Ecology, 100(10),
e02803.
Sazatornil, F. D., More, M., Benitez-Vieyra, S., Cocucci, A. A., Kitching,
I. J., Schlumpberger, B. O., & Amorim, F. W. (2016). Beyond neutral
and forbidden links: Morphological matches and the assembly of mu-
tualistic hawkmoth–plant networks. Journal of Animal Ecology, 85(6),
1586–1594.
Sazima, M., & Sazima, I. (1975). Quiropterofilia em Lafoensia pacari St.
Hil. (Lythraceae), na Serra do Cipó, Minas Gerais. Ciência e Cultura,
27(4), 405–416.
Sritongchuay, T., Hughes, A . C., & Bumrungsri, S. (2019). The role of bats
in pollination networks is influenced by landscape structure. Global
Ecology and Conservation, 20, e 00702. https://doi.org/10.1016/j.
gecco.2019.e00702
Stewart, A . B., & Dudash, M. R. (2017). Flower-visiting bat species con-
tribute unequally toward agricultural pollination ecosystem ser-
vices in southern Thailand. Biotropica, 49(2), 239–248. https://doi.
org /10.1111/btp.12401
Tripp, E. A., & Manos, P. S. (2008). Is floral specialization an evo-
lutionary dead-end? Pollination system transitions in Ruellia
(Acanthaceae). Evolution: International Journal of Organic Evolution,
62(7), 1712–1737.
Vázquez, D. P., Melián, C . J., Williams, N. M., Blüthgen, N., Krasnov, B. R.,
& Poulin, R. (2007). Species abundance and asymmetric interaction
strength in ecological networks. Oikos, 116(7), 1120–1127. https://
doi.org/10.1111/j. 0030-1299.2007.15828.x
Vázquez, D. P., Morris, W. F., & Jordano, P. (2005). Interaction fre-
quency as a surrogate for the total effect of animal mutual-
ists on plants. Ecology Letters, 8(10), 1088–1094. ht tps://doi.
org /10.1111/j.1461-0248. 20 05.0 0810. x
Vizentin-Bugoni, J., Maruyama, P. K., Debastiani, V. J., Duarte, L. D. S.,
Dalsgaard, B., & Sazima, M. (2016). Influences of sampling effort on
detected patterns and structuring processes of a Neotropical plant–
hummingbird network. Journal of Animal Ecology, 85(1), 262–272.
https://doi.or g/10.1111/1365-2656 .12459
Voigt, C . C., Kelm, D. H., Bradley, B. J., & Ortman, S. (2009). Dietary anal-
ysis of plant-visiting bats. In T. H. Kunz, & S. Parsons (Eds.), Ecological
and behavioral methods for the study of bats 2nd, (pp. 593–609). Johns
Hopkins University Press.
Winter, Y., & von Helversen, O. (2003). Operational tongue length in
phyllostomid nectar-feeding bats. Journal of Mammalogy, 84(3), 886–
896. https://doi.org/10.1644/BWG-032
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Supporting Information section.
How to cite this article: Queiroz JA, Diniz UM, Vázquez DP,
et al. Bats and hawkmoths form mixed modules with
flowering plants in a nocturnal interaction network.
Biotropica. 2020;00:1–12. https://doi.org/10.1111/btp.12902
