ArticlePDF Available

Vertically stratified interactions of nectarivores and nectar-inhabiting bacteria in a liana flowering across forest strata

Authors:

Abstract

Premise: Vertical stratification is a key feature of tropical forests and plant-frugivore interactions. However, it is unclear whether equally strong patterns of vertical stratification exist for plant-nectarivore interactions and, if so, which factors drive these patterns. Further, nectar-inhabiting bacteria, acting as "hidden players" in plant-nectarivore interactions, might be vertically stratified, either in response to differences among strata in microenvironmental conditions or to the nectarivore community serving as vectors. Methods: We observed visitations by a diverse nectarivore community to the liana Marcgravia longifolia in a Peruvian rainforest and characterized diversity and community composition of nectar-inhabiting bacteria. Unlike most other plants, M. longifolia produces inflorescences across forest strata, enabling us to study effects of vertical stratification on plant-nectarivore interactions without confounding effects of plant species and stratum. Results: A significantly higher number of visits were by nectarivorous bats and hummingbirds in the midstory than in the understory and canopy, and the visits were strongly correlated to flower availability and nectar quantity and quality. Trochiline hummingbirds foraged across all strata, whereas hermits remained in the lower strata. The Shannon diversity index for nectar-inhabiting bacterial communities was highest in the midstory. Conclusions: Our findings suggest that vertical niche differentiation in plant-nectarivore interactions seems to be partly driven by resource abundance, but other factors such as species-specific preferences of hummingbirds, likely caused by competition, play an important role. We conclude that vertical stratification is an important driver of a species' interaction niche highlighting its role for promoting biodiversity and ecosystem functioning.
Received: 7 September 2023
|
Accepted: 29 January 2024
DOI: 10.1002/ajb2.16303
RESEARCH ARTICLE
Vertically stratied interactions of nectarivores and
nectarinhabiting bacteria in a liana owering across
forest strata*
Sarina Thiel
1
|Malika Gottstein
2
|Eckhard W. Heymann
3
|Jana Kroszewski
1
|
Narges Lieker
1
|Ney Shahuano Tello
4
|Marco Tschapka
5,6
|Robert R. Junker
7
|
Katrin Heer
2
1
Department of Biology, Conservation Ecology,
PhilippsUniversität Marburg, KarlvonFrisch
Str, 8, Marburg, Germany
2
Eva MayrStihl Professorship for Forest Genetics,
AlbertLudwigsUniversität Freiburg, Bertoldstr.
17, Freiburg, Germany
3
Verhaltensökologie & Soziobiologie, Deutsches
Primatenzentrum LeibnizInstitut für
Primatenforschung, Kellnerweg 4, Göttingen,
Germany
4
Estación Biológica Quebrada Blanco, Loreto,
Río Tahuayo, Peru
5
Institute of Evolutionary Ecology and
Conservation Genomics, University of Ulm,
Albert Einstein Allee 11, Ulm, Germany
6
Smithsonian Tropical Research Institute,
Apartado, 084303092, Balboa, Ancon, Republic
of Panama
7
Evolutionary Ecology of Plants, Department of
Biology, PhilippsUniversität Marburg, Karlvon
FrischStr. 8, Marburg, Germany
Correspondence
Sarina Thiel, Department of Biology,
Conservation Ecology, PhilippsUniversität
Marburg, KarlvonFrischStr. 8, Marburg,
Germany.
Email: sarina.thiel@googlemail.com
Katrin Heer, Eva MayrStihl Professorship for
Forest Genetics, AlbertLudwigsUniversität
Freiburg, Bertoldstr. 17, Freiburg, Germany.
Email: katrin.heer@forgen.uni-freiburg.de
Abstract
Premise: Vertical stratication is a key feature of tropical forests and plantfrugivore
interactions. However, it is unclear whether equally strong patterns of vertical stratication
exist for plantnectarivore interactions and, if so, which factors drive these patterns.
Further, nectarinhabiting bacteria, acting as hidden playersin plantnectarivore
interactions, might be vertically stratied, either in response to dierences among strata
in microenvironmental conditions or to the nectarivore community serving as vectors.
Methods: We observed visitations by a diverse nectarivore community to the liana
Marcgravia longifolia in a Peruvian rainforest and characterized diversity and community
composition of nectarinhabiting bacteria. Unlike most other plants, M. longifolia produces
inorescences across forest strata, enabling us to study eects of vertical stratication on
plantnectarivore interactions without confounding eects of plant species and stratum.
Results: Asignicantly higher number of visits were by nectarivorous bats and humming-
birds in the midstory than in the understory and canopy, and the visits were strongly
correlated to ower availability and nectar quantity and quality. Trochiline hummingbirds
foraged across all strata, whereas hermits remained in the lower strata. The Shannon
diversity index for nectarinhabiting bacterial communities was highest in the midstory.
Conclusions: Our ndings suggest that vertical niche dierentiation in
plantnectarivore interactions seems to be partly driven by resource abundance,
but other factors such as speciesspecic preferences of hummingbirds, likely caused
by competition, play an important role. We conclude that vertical stratication is an
important driver of a speciesinteraction niche highlighting its role for promoting
biodiversity and ecosystem functioning.
KEYWORDS
Bats, hummingbirds, Marcgraviaceae, nectar traits, nectarivores, nectarinhabiting bacteria, plantanimal
interactions, rain forest
Vertical stratication is a widespread phenomenon in plant
and animal communities and a key factor for structuring
biodiversity, particularly in tropical forests (Basset et al., 2003;
Chmel et al., 2016;Thieletal.,2021). Plants and animals
occupy a variety of niches along the vertical forest gradient
(Allee et al., 1949; Richards, 1952;Smith,1973;Bongers,2001).
Patterns of vertical stratication for nectarivores and frugi-
vores have been observed in terms of species abundance,
Am J Bot. 2024;111:e16303. wileyonlinelibrary.com/journal/AJB
|
1of18
https://doi.org/10.1002/ajb2.16303
This is an open access article under the terms of the Creative Commons AttributionNonCommercial License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited and is not used for commercial purposes.
© 2024 The Authors. American Journal of Botany published by Wiley Periodicals LLC on behalf of Botanical Society of America.
*We dedicate this paper to the memory of our colleague and friend Dr. Stefan Dressler, an international leading expert for the Marcgraviaceae, who died on 14 September 2023.
richness, and community composition (BuchananSmith
et al., 2000; Kalko and Handley, 2001;Chmeletal.,2016;
Thiel et al., 2021). Thus, this vertical structure also aects
associated ecological processes such as pollination and seed
dispersal (Howe and Smallwood, 1982;Fleming,1993;
Jordano, 2000; Fleming and Kress, 2013).
In tropical rainforests, vertebrates play crucial roles as
pollinators and seed dispersers (Fleming and Kress, 2013). For
instance, nectarivorous bats and hummingbirds are among the
key pollinators whereas frugivorous bats, birds, and primates are
key seed dispersers (Fleming and Muchhala, 2008; Lobova
et al., 2009). However, studies focusing on the actual
interactions among taxa across the vertical gradient are scarce.
For plantfrugivore interactions, for instance, it was shown that
networks dier profoundly among strata in terms of mutual
specialization and interaction frequency and community
composition of frugivores (Shanahan and Compton, 2001;
Schleuning et al., 2011; Thiel et al., 2023). Few studies have
examined the vertical stratication of interactions among plants
and nectarivores although there are known dierences in strata
use among taxa. In tropical lowland forests, hermit humming-
birds (subfamily Phaethornithinae) are primarily understory
foragers, whereas trochiline hummingbirds (subfamily Trochi-
linae) forage in the canopy and in the understory (Fleming and
Kress, 2013). This separation in vertical space (Feinsinger and
Colwell, 1978;Stiles,1981;Flemingetal.,2005) might have been
driven by the high level of interspecic competition among
hummingbirds in tropical lowland forests and the subsequent
specialization on diering oral resources (Maglianesi
et al., 2015). For nectarivorous phyllostomid bats, such a clear
distinction between understoryand canopyfeeding bats does
not seem to exist. Some studies indicate that bats prefer to feed
on owers higher up due to better accessibility and conspicuity
(Diniz et al., 2019; Kobayashi et al., 2020), whereas others report
bats feeding on plants owering in the understory (Czenze and
Thurley, 2021;Amorimetal.,2023). Generally, it seems that
most neotropical, nectarivorous bats are rather exible and use
all levels of the forest for foraging (Kalko and Handley, 2001;
Fleming et al., 2005).
Plantnectarivore interactions are determined by a series of
ecological factors and evolutionary processes (Carnicer
et al., 2009). One crucial ecological factor is resource availability
(GonzálezCastro et al., 2012). Hummingbirds and nectarivor-
ous bats are highly dependent on nectar resources with a high
sugar content (Stiles, 1981; Temeles et al., 2005;Suarezand
Welch, 2017), and competition for suitable nectar resources
plays an important role in structuring their community
organization (Feinsinger and Colwell, 1978; Kalko and
Handley, 2001). Nectar components such as sugars, amino
acids, lipids, and other nutrients (Carter et al., 2006), inuence
the attractiveness of nectar for nectarivores. However, the
relationship between nectar and pollinators is far more complex
than originally assumed. For instance, by adjusting nectar
attributes and nectar secretion of already pollinated owers,
plants manipulate pollinators to rather visit unpollinated owers
with high pollen availability (Pyke et al., 2020;DomingoMellos
et al., 2023). Moreover, nectar harbors an abundant and diverse
microbial community. These microorganisms might be impor-
tanthiddenplayersinplantnectarivore interactions as their
metabolism can profoundly alter nectar chemistry and thus,
inuence nectar consumption by nectarivores (Vannette
et al., 2013). So far, most studies on the nectar microbiome
concentrated on nectarinhabiting fungi and yeast (Lachance
et al., 2001a;BryschHerzberg, 2004; Herrera et al., 2008;de
Vega et al., 2009;Pozoetal.,2009,2011), even though another
group of microbes, bacteria, are also frequently found in nectar
with a high abundance and diversity (ÁlvarezPérez et al., 2012;
Fridman et al., 2012; Junker and Keller, 2015;Lievensetal.,2015;
Vannette, 2020; Gaube et al., 2021). Bacterial metabolism results
in a decline in total sugar concentration (Herrera et al., 2008;de
Vega et al., 2009), changes the sugar and amino acid
composition (Herrera et al., 2008; Canto and Herrera, 2012),
and increases nectar temperature (Herrera and Pozo, 2010).
Further, they emit volatiles that can modify ower and nectar
scent (Golonka et al., 2014;Pozoetal.,2014;Helletsgruber
et al., 2017;Reringetal.,2018). As these nectar traits are key
mediators of interactions between plants and nectarivores, such
changes in physiochemical properties of the oral nectar can
alter the attractiveness of a given ower to pollinators (Herrera
et al., 2008,2009; Herrera and Pozo, 2010; Vannette et al., 2013;
Junker et al., 2014;Lievensetal.,2015;Stevensonetal.,2017).
These bacterial eectsmayevenbestrongerthanthosebyyeast
(Vannette et al., 2013). At the same time, nectarivores
themselves are important vectors for the dispersal of nectar
inhabiting bacteria among plants (Sandhu and Waraich, 1985;
de Vega et al., 2009; Vannette et al., 2013). Thus, nectarivores
and microenvironmental conditions such as temperature,
rainfall, and vegetation density inuence the structure and
diversity of nectarinhabiting bacteria communities, which in
turn aect nectar characteristics (SamuniBlank et al., 2014;
Sharaby et al., 2020;Bogoetal.,2021). If nectarivores and
microenvironmental conditions are vertically stratied
(Shaw, 2004; Fleming and Kress, 2013), it is plausible that
there is vertical stratication of nectarinhabiting bacterial
communities, which in turn might enhance vertical dierences
of plantnectarivore interactions.
Although there is extensive information about plant
nectarivore interactions, to our knowledge, all studies investi-
gating the foraging behavior of nectarivores have focused on
their specialization on plant species, which presented owers
within a single stratum, implying that species and stratum are
partially confounding variables. Yet, some plant species, most of
them cauliorous, produce owers and fruits over more than
one stratum. The neotropical liana Marcgravia longifolia
(Marcgraviaceae) is an ideal study organism to investigate the
vertical stratication in plantanimal interactions because it
provides inorescences and infructescences from ground level
to the canopy. We already showed that frugivorous birds
feeding on M. longifolia fruits stayed within their preferred
vertical niche, even though the same resource was available
across strata (Thiel et al., 2023). Based on these results, we
concluded that vertical stratication is driven by inherent
2of18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
preferences of avian frugivores for specic strata rather than by
dierences in the composition of plant communities. In the
frugivore community, species dependence on fruit in their diet
and speciesspecictraitssuchasbodysizeandwingshapein
combination with microenvironmental characteristics such as
vegetation density and canopy cover played an important role
(Thiel et al., 2023).Hereweusedthesamesystemtotest
whether nectarivores also prefer a certain vertical niche for
foraging or whether an attractive resource that is available across
strata drives nectarivorous hummingbirds and bats toward
foraging across the whole vertical gradient. We further
investigated whether the nectarinhabiting bacterial communi-
ties of M. longifolia have any vertical stratication.
So far, the few available studies did not unequivocally test
whether the use of vertical space by nectarivores is driven by
dierential resource availability or is due to speciesspecic
preferences for a certain vertical niche. Our unique study system
will allow us to test these possibilities in the presence of the
same resources along the entire vertical gradient. Specically, for
the diurnal and nocturnal nectarivores, we consider two
contrasting hypotheses (here: H1 and H2) as relevant.
According to H1 nectarivores preferentially forage in a distinct
vertical niche based on speciesspecicpreferences.Ifsucha
preference exists, the number of visits, species diversity, and
community composition are expected to dier among strata in
our study system. In contrast, according to H2, resource
quantity and quality are decisive for determining the vertical
foraging niche. In that case, we would expect that nectarivores
forage across strata with an increasing number of visits with
increasing inorescence abundance, nectar quantity, and sugar
concentration. Because hummingbirds have been shown to use
diering foraging strategies to reduce interspeciccompetition
and because we also found vertical patterns for frugivorous
birds, we assumed that for hummingbirds both H1 and H2
provide realistic scenarios. For nectarivorous bats, on the other
hand,weexpectH2tobemorelikelyasbatshavebeenshown
to be more exible and to use all vertical strata for foraging.
Finally, vertically stratied foraging behavior of nectarivores,
microclimatic conditions, and nectar properties inuence
communities of nectarinhabiting bacteria. Thus, we hypothe-
sized that their diversity and composition is also vertically
stratied (H3). In our study system, we expected that
microenvironmental dierences among strata are a strong
determinant of the vertical stratication of bacteria communi-
ties, which might either be fortied or homogenized depending
on how frequently nectarivores forage across strata. Since we
assumed that nectar properties only dier negligibly among
strata, they should not have an inuence in our speciccase.All
hypotheses are tested against H0 of no dierences among strata.
To test our hypotheses and expectations, we investigated
ower quality and nectarivoreforaging behavior across the
vertical gradient by collecting microenvironmental data,
sampling nectar from M. longifolia individuals, and analysing
nectar quantity and sugar concentration across strata and
during anthesis and the 24 hcycle. We further recorded visits
of hummingbirds and nectarivorous bats and diversity and
community composition of the nectarinhabiting bacteria in
M. longifolia individuals across forest strata (understory,
midstory, canopy).
MATERIALS AND METHODS
Study site
The study was conducted at the Estación Biológica Quebrada
Blanco in northeastern Peruvian Amazonia (4°21S 73°09W;
EBQB, Loreto, Peru) on high ground terra rme rainforest
(bosque de altura; following Encarnación [1985]), inter-
spersed with swampy areas. Annual precipitation is ca.
3000 mm, with DecemberMay the wettest months and
JulyAugust the driest (Lüeetal.,2018). Mean monthly
temperatures in the area range between 25° and 27°C
(Klingbeil and Willig, 2008). Further details of the study site
have been described by Heymann et al. (2021) and Heymann
and Tirado Herrera (2021).
Plant species
Marcgravia longifolia (Marcgraviaceae) is a woody liana
species only known from western Amazonia (Tropicos, 2020).
It produces long pedunculate and agelliorous inores-
cences arising from the unbranched stem of the liana all the
way from ground level to the canopy, which is an extremely
rare phenomenon within the Marcgraviaceae and for plants
in general (Appendices S1,S2).
Data collection
Selection of Marcgravia longifolia individuals
We selected 29 of the 100 known M. longifolia individuals at the
study site for further observations of nectarivores and for nectar
sampling based on resource availability (sucient number of
inorescences) and visibility, and accessibility of the host tree
with climbing equipment. Data on owering M. longifolia
individuals were collected in October 2017, from August to
September 2018, and from July to September 2019.
Classication of height and microenvironmental
variables
We determined host tree height for each selected individual and
additionally characterized canopy cover along the vertical
gradient by collecting diagonal hemispherical color photographs
to assess canopy cover. Additionally, we installed data loggers in
11 M. longifolia individuals in dierent heights to collect data on
temperature and light intensity (Appendix S1). We classied
three forest strata (understory, midstory, and canopy) in relation
to the vertical distribution of foliage and density of the
surrounding vegetation. We found that vegetation density was
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
3of18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
clearly distinct among these strata, and we assume that
nectarivores rather orient themselves along the vegetation
structure than based on absolute height (McCaig et al., 2020).
The understory was classied according to the height of the
dense surrounding shrub and palm tree layer (0 m until
between 3 m and 10 m). For the classication of the canopy
stratum, the height of the rst canopy branch of the host tree
and the height of the surrounding canopies was decisive (from
between9mand22mtobetween12mand32m).The
midstory was then dened as the space between the understory
and the canopy, where vegetation density was lower (from
between 3 m and 10 m to between 9 m and 22 m; Appendix S3).
Nectar sampling across strata
To characterize diversity and community composition
of bacteria communities, we collected nectar from one
inorescence per stratum from ve plants of M. long-
ifolia in 2018. For each individual, samples were
collected on the same day. We selected inorescences
in which some owers had already shed their calyptra to
ensure that the inorescences were anthetic and had
already produced nectar (Appendix S1).
To assess dierences in nectar production, sugar concen-
tration, and pH values among strata, we collected nectar from
one inorescence per stratum from seven M. longifolia
individuals in 2019. We selected inorescences within anthesis.
In each M. longifolia individual, samples were collected on the
samedayandalwaysinthelatemorningorearlyafternoon,
respectively, when dierences among strata in terms of
temperature and light intensity were greatest (see Results,
Microenvironmental variables). We measured nectar volume,
sugar concentrations, and pH. We then calculated mean nectar
volume as the sum of nectar from all nectaries of one
inorescence divided by the number of nectaries. Mean pH and
mean sugar concentration were calculated as the mean of the
values for all sampled nectaries from one inorescence. Lastly,
mean sugar quantity was calculated by multiplying mean nectar
quantity and mean sugar concentration (Appendix S1).
Nectar sampling during anthesis and over 24 h
In 2019, we selected a total of 13 inorescences from 10 M.
longifolia individuals, which were accessible from the
ground, to examine nectar volume, sugar concentration,
and pH during anthesis. Nectar measurements started with
the onset of owering until nectaries were shed (due to time
constraints, some measurements had to be stopped earlier).
Measurements were always done at the same time of day.
To determine whether nectar quantity and sugar concen-
tration show a variation over the 24h cycle, we selected a total
of nine inorescences from six M. longifolia individuals at
ground level, which were approximately at the same stage of
anthesis. Measurements were made at 2hintervalsfor24h.For
each inorescence, we determined the sampling height, the
number of nectaries, the total number of owers, and the
number of owers that had already shed their calyptra
(Appendix S1).
Identity and visitation rate of diurnal nectar
consumers
To record nectarivorous species feeding on M. longifolia
during the day, four researchers equipped with binoculars
(10 × 42 mm) recorded animal visits to 26 owering M.
longifolia individuals in the morning (06:00 to 11:00 hours)
and in the afternoon (12:00 to 17:00 hours; Appendix S4).
The number of visits by each hummingbird species was
calculated per stratum reachM. longifolia individual and
summed across years.
Furthermore, by selecting one to two nectarproducing
inorescences per stratum (indicated by hummingbird
visits), for all 26 M. longifolia individuals, and counting
the number of visits at those inorescences, we determined
the visitation rate of diurnal nectarivores in a way that is
comparable to the nocturnal visitation rate derived from the
wildlife cameras (see below; we did not use the wildlife
cameras to monitor diurnal visitation because humming-
birds did not trigger them reliably, probably due to the
insulating characteristic of the hummingbirdsplumage and
the small dierence between hummingbird body and
ambient temperature). For all 26 M. longifolia individuals,
the number of all inorescences irrespective of their
phenology state was estimated for each height class of
4 m, on each observation day, and then assigned to strata.
We calculated mean number of inorescences per stratum
of each M. longifolia individual across observation days. To
estimate the availability of alternative ower resources, close
to the observed M. longifolia individuals, we searched for
other plant species bearing owers in a radius of 15 m
around the M. longifolia individuals (Appendix S1).
Identity and visitation of nocturnal nectarivores
To determine the visitation rate of nocturnal nectarivores,
we xed automatic wildlife cameras (HyperFire 2TM HF2X;
Reconyx, Holmen, WI, USA; with additional IRLED) on
the liana stem close to inorescences during anthesis at
heights between 0.5 and 23 m of 14 owering M. longifolia
individuals (Appendix S1). We further monitored the
inorescences sampled during anthesis with the wildlife
cameras, enabling us to directly observe nocturnal visita-
tions by nectarivores to these inorescences during anthesis
and to directly correlate visitation to nectar volume and
sugar concentration. We counted each individual animal
that contacted the monitored inorescences and calculated
the number of visits per M. longifolia individuals and
summed the visits to the individual across years.
To identify the nectarivorous bat species visiting ower-
ing M. longifolia individuals, we set up nocturnal mist nets in
4of18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
heights from 0.2 to 7.6 m as close as possible to 10 owering
M. longifolia individuals and at control sites. We collected
pollen samples from nectarivorous bats using a small block of
fuchsinstained gelatine (Appendix S1).
Metabarcoding of bacteria communities in the
nectar
In the lab, the sample tubes containing the nectar were stored at
20°C until extraction. We extracted microbial DNA from
nectar samples following the protocol of the ZymoBIOMICS
DNA Miniprep Kit (D4300; Zymo Research, Irvine, CA, USA).
ThemicrobiomeforeachsamplewasanalyzedbyEurons
Genomics (Ebersberg, Germany) using the company's standard
procedure. Sequencing was done using Illumina MiSeq, and the
sequenced regions were V3V4 region of the 16 S rRNA gene to
identify bacterial operational taxonomic units (OTUs) using the
standard procedure by the service INVIEW Microbiome
Proling 3.0 with MiSeq (Eurons Genomics, Luxembourg)
(for detailed methods, see Junker et al., 2020). Abundances of
bacterial OTUs were normalized using lineagespeciccopy
numbers of the relevant marker genes to improve estimates
(Angly et al., 2014). The Shannon diversity index (Shannon,
1948) was calculated based on the OTU composition (without
CSS normalization) after rarefying the data to the minimum
number of reads (N= 17,502) available in the samples
(repeats = 999) (Appendix S1).
Network analysis
We built an interaction matrix between hummingbird
species and the three forest strata (understory, midstory,
and canopy), summed across M. longifolia individuals. This
way, we calculated interaction frequencies of each hum-
mingbird species for each stratum (number of visits of
feeding animals within that stratum). Following traditional
network analyses, the interaction matrix was analyzed using
the R package bipartite (Dormann et al., 2009) in R version
4.2.2 (R Core Team, 2019). To build the matrix, we used a
quantitative matrix of interactions between hummingbirds
and strata, in which the nodes represent hummingbird
species or plant strata, respectively, and the links represent
the interaction frequency between them. The total fre-
quency of a hummingbird species was dened as the
number of visits to all strata, whereas the total frequency
from the perspective of a particular stratum was given as the
number of all hummingbird visits to this stratum (Blüthgen
et al., 2007).
To characterize hummingbirdforaging behavior across
strata, we rst assessed the degree of specialization for each
species by calculating Pielou's evenness index for visited strata
per species. Values of 0 indicate specialist species, values of 1
indicate generalist species. Second, we calculated the Shannon
diversity index for hummingbird interactions in each stratum
(Dormann, 2011).
Statistical analyses
Microenvironmental variables, resource quantity
and quality
To examine the factors that might inuence the foraging
behavior of nectarivores, we analyzed whether micro-
environmental conditions and resource quantity and quality
diered among strata. First, we compared canopy closure
and number of inorescences among strata by calculating
linear mixed eect models with canopy closure or number
of inorescences as response variables, stratum as explana-
tory variable, and IDs of individual lianas as random factors
(Table 1). Residuals of canopy closure and number of
inorescences were normally distributed. Then, for those
individuals that were sampled for nectar, we compared
nectar attributes among strata by calculating linear mixed
eect models with mean nectar volume, mean sugar
concentration, mean sugar volume, or mean pH value as
response variables; stratum, total number of owers per
inorescence, and percentage of open owers as explanatory
variables; and IDs of individual lianas as random factors
(Table 1). Further, to identify whether nectar production
changed during anthesis, we calculated linear mixed eect
models with mean nectar volume, mean sugar concentra-
tion, or mean pH value as response variables; days of
anthesis, total number of owers per inorescence, and
percentage of open owers as explanatory variables; and IDs
of individual lianas and of inorescences nested in IDs of
individual lianas as random factors (Table 1). To test for
dierences over the 24h cycle, we ran a WatsonWilliams
test for circular distributions to analyze whether there is a
nonuniformity in mean nectar quantity and mean sugar
concentration over the 24h cycle, pooled across all sampled
M. longifolia individuals.
Foraging behavior of nocturnal and diurnal
nectarivores
To test whether hummingbirds prefer a certain vertical niche
for foraging (Hypothesis 1 (H1)), we rst analyzed whether the
degree of specialization diers among hermits and trochilines.
For this purpose, we calculated a linear mixed eect model
with Pilou's evenness index as response variable, hummingbird
subfamily as explanatory variable, and species identity as
random factor (Table 1). Then, we compared the zeroinated
count data of hummingbird visitation among strata by
running a generalized linear mixed eect model with number
of hummingbird visits as response variable; the interaction of
stratum, subfamily and number of inorescences as explana-
tory variables; and IDs of individual lianas and species identity
as random factors. We also added a zeroination term for
stratum, an oset for observation hours (to account for
diering number of hours of observation among M. longifolia
individuals) and tted the model with a negative binomial
distribution and a logit link (Table 1). For the analyses,
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
5of18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 1 Results of the linear and generalized linear mixed eect models examining dierences in microenvironmental variables and resource quantity and
quality among strata, days of anthesis, and inorescence characteristics. Further models tested the eect of stratum, hummingbird subfamily or the interaction of strata
and hummingbird subfamilies/nocturnal and diurnal taxa on hummingbird visitation, Pilou's evenness index, and nocturnal and diurnal visitation rate. The
Marcgravia longifolia individual, hummingbird species identity, or the individual inorescences, respectively, were included as random factors. Pwas derived using the
glmmTMB() function.
Model Variable χ
2
df P
Microenvironmental variables
Canopy closure ~ Stratum + (1|Marcgravia ID) Stratum 41.82 2 <0.0001
Resource quantity and quality
Number of inorescences ~ Stratum + (1|
Marcgravia ID)
Stratum 200.39 2 <0.0001
Mean nectar volume ~ Stratum + Percentage open
owers + Total number of owers + (1|
Marcgravia ID)
Stratum 0.39 2 0.821
Percentage open owers 6.098 1 0.014
Total number of owers 1.376 1 0.241
Mean sugar concentration ~ Stratum + Percentage
open owers + Total number of owers + (1|
Marcgravia ID)
Stratum 2.126 2 0.345
Percentage open owers 29.631 1 <0.0001
Total number of owers 8.163 1 0.004
Mean sugar volume ~ Stratum + Percentage open
owers + Total number of owers (1|
Marcgravia ID)
Stratum 0.444 2 0.801
Percentage open owers 14.749 1 0.0001
Total number of owers 1.479 1 0.224
Mean pH value ~ Stratum + Percentage open
owers + Total number of owers + (1|
Marcgravia ID)
Stratum 3.787 2 0.151
Percentage open owers 0.186 1 0.667
Total number of owers 0.13 1 0.719
Mean nectar volume ~ Day of anthesis + Percentage
open owers + Total number of owers + (1|
Marcgravia ID/Inorescence ID)
Day of anthesis 16.84 2 0.0002
Percentage open owers 8.19 1 0.004
Total number of owers 4.485 1 0.027
Mean sugar concentration ~ Day of
anthesis + Percentage open owers + Total
number of owers + (1|Marcgravia ID/
Inorescence ID)
Day of anthesis 58.136 2 <0.0001
Percentage open owers 15.168 1 <0.0001
Total number of owers 0.859 1 0.354
Mean pH ~ Day of anthesis + Percentage open
owers + Total number of owers + (1|Marcgravia
ID/Inorescence ID)
Day of anthesis 2.785 2 0.248
Percentage open owers 0.127 1 0.721
Total number of owers 1.013 1 0.314
Hummingbird visitation
Number of hummingbird
visits ~ Stratum × Subfamily + Number of
inorescences + (1| Species) + (1|Marcgravia
ID) + oset = (Hours), zi = ~ Stratum,
family = nbinom2
Stratum × Subfamily 8.102 2 0.017
Stratum 6.201 2 0.045
Subfamily 0.134 1 0.714
Number of inorescences 0.378 1 0.539
Pilou's evenness index
Pilou's evenness index ~ Subfamily + (1| Species) Subfamily 5.97 1 0.015
Nocturnal and diurnal visitation rate
Number of visits ~ Stratum × Taxon + Number of
inorescences + (1|Marcgravia ID) + oset
(Hours), zi = Stratum, family = Poisson
Stratum * Taxon 120.06 6 <0.0001
Stratum 918.46 2 <0.0001
6of18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
number of inorescences and observation hours were log
transformed to approximate normal distribution of residuals.
In all models, we used a contrast to compare the estimated
marginal means of the response variables among strata or
among strata and diering hummingbird subfamilies
(Table 2). To quantify the dierences in the hummingbird
community among forest strata, we computed pairwise
BrayCurtis distances among strata and analyzed dierences
in community composition among strata using a MANOVA
approach on our interaction matrix. We tested signicance by
permuting the raw data (1000 permutations) using the
function adonis() in R package vegan 2.56(Oksanen
et al., 2019).
Totestwhetherresourcequantityandqualityaredecisive
for determining the vertical foraging niche (H2), we compared
the visitation rate of nocturnal and diurnal nectarivores among
strata by calculating a generalized linear mixed eect model
with the number of visits per taxon as response variable, the
interaction of stratum and taxon (bat, bird, marsupial, or
moth) and the number of inorescences as explanatory
variables, and IDs of individual lianas as random factors. We
also added a zeroination term for stratum, an oset for
observation hours, and tted the model with a Poisson
distribution and a logit link (Table 1). For the analyses,
number of inorescences was logtransformed to approximate
normal distribution of residuals. Further, the visitation rate of
nocturnal nectarivores during the anthesis was compared by
calculating a generalized linear mixed eect model with
number of visits as response variable; day of anthesis, mean
nectar quantity, and mean sugar concentration as explanatory
variables; and IDs of individual lianas and of inorescences
nested in IDs of individual lianas as random factors. We also
added an oset for observation hours and tted the model
with a Poisson distribution and a logit link (Table 1). In all
models, a contrast was used to compare among strata or days
of anthesis (Table 2; Appendix S5).
Diversity and community composition of nectar
inhabiting bacteria
To test whether the diversity and community composition of
nectarinhabiting bacteria dier among strata (H3), we rst
performed a nonmetric multidimensional scaling (NMDS)
based on the BrayCurtis distances between nectar samples
using the function vegdist() in R package vegan 2.57
(Oksanen et al., 2019) to compare the community composition
among strata. Before the NMDS, we performed a cumulative
sum scaling (CSS) normalization in R package metagenome-
Seq. 1.28.2 (Paulson et al., 2013) on the read count data to
account for dierences in sequencing depth among samples.
On top of ordination, we tted microenvironmental vectors
and factors (plant individual, height, nectar volume, number of
owers during anthesis, total number of owers, and
proportion of owering plants) to test for their eects on
bacterial composition using the function envt() in R package
vegan 2.57(Oksanenetal.,2019). Next, we compared the
diversity of nectarinhabiting bacteria among strata. To
account for dierent numbers of sequencing reads per samples
we rst rareed (999 repeats) the unscaled data set (raw
number of reads) to the minimum number of reads available
in the samples (N = 17,502) and then calculated the Shannon
diversity index for each sample based on the OTU assignment
of each read using the R package rtk 0.2.6.1 (Saary et al., 2017).
Diversity was correlated to height using a Spearman correla-
tion analysis. Visual inspection of the relationship between
height and bacterial diversity suggested a nonmonotonic
relationship with highest diversity values in the midstory.
Accordingly, we used the function qad() in R package qad
1.0.0 (Junker et al., 2021;Griessenbergeretal.,2022)totestfor
a nonmonotonic relationship. The package qad estimates
scaleinvariantdirected and asymmetric dependence of
bivariate distributions with no underlying assumptions on
the distribution of the data and the function type.
Linear and generalized linear mixed models were calculated
using the function glmmTMB() in R package glmmTMB 1.0.2.1
(Brooks et al., 2020). We used the function Anova() in R
package car3.010 (Fox and Weisberg, 2019)forWaldχ
2
tests
and determined contrast comparisons with the function
emmeans() in R package emmeans 1.4.7 (Lenth et al., 2020).
RESULTS
Microenvironmental variables
Microenvironmental conditions diered among strata. Canopy
closure signicantly decreased from the understory to the
canopy (Tables 1and 2; Appendix S5). Even though we could
not compare temperature and light intensity measurements
TABLE 1 (Continued)
Model Variable χ
2
df P
Taxon 4302.97 3 <0.0001
Number of inorescences 15.789 1 <0.0001
Number of visits ~ Day of anthesis + mean nectar
quantity + mean sugar concentration + (1|
Marcgravia ID/Inorescence ID) + oset (Hours),
family = Poisson
Day of anthesis 1318.58 2 <0.0001
Mean nectar quantity 28.81 1 0.784
Mean sugar concentration 0.08 1 <0.0001
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
7of18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TABLE 2 Standardized estimates, standard errors (SE), and Pvalues for all models derived from the emmeans() function. These contrast comparisons
were conducted for the glmmTMB() models in Table 1examining dierences among strata, among strata and hummingbird subfamilies/taxa, or among strata
and inorescence characteristics, respectively, on microenvironmental variables, day of anthesis, hummingbird visitation, or nocturnal and diurnal visitation.
The Marcgravia longifolia individual, the individual inorescences, or hummingbird species identity, respectively, were included as random factors.
Response variable Contrast Standardized estimates SE P
Microenvironmental variables
Canopy closure High Low 5.01 0.81506 <0.0001
High Middle 2.15 0.808 0.009
Low Middle 2.87 0.679 0.0001
Number of inorescences High Low 56.2 4.04 <0.0001
High Middle 29.1 4.04 <0.0001
Low Middle 27.1 3.33 <0.0001
Day of anthesis
Nectar quantity Before After 0.005 0.017 0.774
During After 0.029 0.011 0.016
Before During 0.033 0.012 0.016
Sugar concentration Before After 3.525 0.85 0.0002
During After 0.873 0.536 0.109
Before During 4.399 0.607 <0.0001
Hummingbird visitation
Hermits Low High 7.175 4.211 0.012
Middle High 3.36 1.945 0.146
Low Middle 2.135 0.624 0.071
Trochilines Low High 1.355 0.615 0.755
Middle High 2.274 0.555 <0.0001
Low Middle 1.064 0.221 0.834
Nocturnal and diurnal visitation
Stratum Low High 1.896 0.177 <0.0001
Middle High 2.299 0.212 <0.0001
Low Middle 0.825 0.056 0.004
Bird Low High 2.688 0.399 <0.0001
Middle High 2.274 0.339 <0.0001
Low Middle 1.182 0.07 0.007
Bat Low High 2.638 0.146 <0.0001
Middle High 4.051 0.203 <0.0001
Low Middle 0.65 0.017 <0.0001
Moth Low High 1.799 0.438 0.018
Middle High 1.812 0.45 0.019
Low Middle 0.993 0.178 0.968
Marsupial Low High 1.014 0.213 0.964
Middle High 1.673 0.355 0.018
Low Middle 0.606 0.113 0.009
8of18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
statisticallyduetoinsucient data, for many individuals, the
temperature and light intensity were, as expected, greater in
the canopy than in the lower strata (Appendix S6). Flowering
individuals from ve plant species grew in the vicinity of the
observed M. longifolia individuals (Appendix S7). However, in
comparison to M. longifolia,theyproducedfewowers and
were thus not further considered in our analysis (N= seven
individuals with <50 owers, N=1 with >50 owers). All
individuals produced their owers high in the canopy.
Resource quantity and quality
Resource quantity diered among strata with more inor-
escences in the under, and midstory than in the canopy
(Tables 1and 2,Figure1A;AppendixS5). Generally, nectar
production by M. longifolia ranged from 0.15 to 0.65 mL
(mean 0.475 ± 0.028 mL) of nectar per inorescence with a pH
between 4 and 9. Mean nectar volume, mean sugar
concentration, and mean pH did not signicantly dier
among strata (Table 1,Figure1; Appendix S5). For mean sugar
volume, we found a nonsignicant trend with a slightly higher
mean sugar volume in the canopy than in the underand
midstory (Table 1,Figure1; Appendix S5). Nectar production
varied temporally with higher mean volume and sugar
concentration during anthesis compared to before or after
anthesis (Tables 1and 2,Figure2A, B; Appendix S5). The pH
remained relatively constant over time (Table 1;AppendixS5).
We also found a circadian variation; mean nectar volume
(WatsonWilliams test: F = 13.35, df = 8, P < 0.0001) and
mean sugar concentration (WatsonWilliams test: F = 6.02,
df = 8, P < 0.0001) was highest during the night (nectar mean:
0.043 ± 0.004; sugar mean: 8.26 ± 0.48) and early morning
(nectar mean: 0.042 ± 0.004; sugar mean: 8.42 ± 0.37;
Figure 2C, D). In all models, mean nectar volume and mean
sugar concentration signicantly increased with the percentage
of open owers (Table 1; Appendix S5).
Identity of nocturnal and diurnal nectarivores
In total, during focal observations when all inorescences of
M. longifolia visited by hummingbirds were considered, we
observed 1146 interactions between 26 owering individuals
and 12 hummingbird species (three hermit and nine
trochiline species, Figure 3; Appendix S8). The most frequent
hummingbird species on owers were the needlebilled
hermit (Phaethornis philippii, N = 344 observations), Gould's
jewelfront (Heliodoxa aurescens, N = 284), and the forktailed
woodnymph (Thalurania furcata, N = 276). During the
observations of visitations to the focal inorescences per
stratum, we observed 1233 visits by hummingbirds.
The wildlife cameras documented 7946 visits by
nectarivorous bats (96.3% of the recorded visits; 2.7 visits
per observation hour), 155 visits by marsupials (1.9%; 0.05
visits per observation hour), and 152 visits by moths (1.8%;
0.05 visits per observation hour) to owering M. longifolia
individuals. All marsupials that could be identied from
wildlife camera photos belonged to the genera Marmosa and
Micoureus (Didelphidae), and the only identiable moth
species (40% of observed moth individuals) was Feigeria
scops (Erebidae).
In front of 10 owering M. longifolia individuals, we
captured 89 bat individuals belonging to 16 species all from
the family Phyllostomidae. Of these, 66 individuals were
nectarivores (74.2%), 20 frugivores (22.5%), and three
animalivores (3.4%; Appendix S7). Among the nectarivor-
ous species, Thomas's nectar bat was the most abundant
species (Hsunycteris thomasi, N = 53 captures), followed by
Choeroniscus minor (N = 8), and the tailed tailless bat
(Anoura caudifer, N = 5). Of 37 pollen samples collected
from H. thomasi, 27 samples contained pollen from
M. longifolia. All three of the pollen samples from C. minor
and two of three from A. caudifer contained M. longifolia
pollen. In contrast, the percentage of nectarivores was much
lower at control sites, where we caught 38 bat individuals, of
TABLE 2 (Continued)
Response variable Contrast Standardized estimates SE P
Taxa Bat Bird 4.44 0.259 <0.0001
Bat Marsupial 19.17 1.604 <0.0001
Bat Moth 28.35 2.683 <0.0001
Bird Marsupial 4.31 0.431 <0.0001
Bird Moth 6.38 0.687 <0.0001
Marsupial Moth 1.48 0.182 0.002
Day of anthesis After Before 21.97 0.311 <0.0001
After During 22.98 0.196 <0.0001
Before During 1.01 0.244 <0.0001
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
9of18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
which 23 were frugivores (60.5%), nine were animalivores
(23.7%), and six were nectarivores (15.8%; Appendix S9).
Foraging behavior of nocturnal and diurnal
nectarivores
In line with H1 (nectarivores prefer a certain vertical niche
for foraging), hummingbird visitation diered among strata
(Table 1, Figure 4; Appendix S5). While hermits foraged with
signicantly fewer visits to the canopy, trochilines foraged
with equal frequency across strata (Tables 1and 2, Figure 4A;
Appendix S5). The diering visitation of hermits and
trochilines among strata was not correlated with number of
inorescences (Table 1). Also, the Pilou's evenness index
diered among hermits and trochilines and indicated greater
specialization for a specic stratum for hermits than for
trochilines (Table 1,Figure4B;AppendixS5). In contradic-
tion to H1, neither the Shannon diversity index (understory:
1.5, midstory: 1.63, canopy: 1.43), nor the hummingbird
community composition diered among strata (adonis:
r
2
=0.033, F= 0.95, P=0.53; Figure 3).
In line with H2 (resource volume and quality are decisive
for determining the vertical foraging niche of nectarivores), the
visitation rate of nocturnal and diurnal nectarivores among
strata was positively correlated with inorescence abundance
(Tables 1and 2,Figure5A;AppendixS5), which were both
higher in the understory than in the canopy. Similarly,
visitation rate was higher during anthesis than before or after
anthesis. Visitation further increased with sugar concentration
(Tables 1and 2,Figure5C, D; Appendix S5). Bats had the
highest visitation rates among all taxa in all strata but foraged
more often in the midstory than in the understory and canopy.
Birds, on the other hand, foraged less frequently in the canopy
(Tables 1and 2,Figure5B;AppendixS5).
Diversity and community composition of nectar
inhabiting bacteria
In contradiction to H3, the bacterial community composition
neither diered among strata nor was it inuenced by any
other of the tested explanatory variables (Figure 6A). The plant
individual,itsheight,oranyower or inorescence character-
istics (nectar volume, number of owers during anthesis, total
number of owers, and proportion of owering plants) did not
aect the composition of bacteria in ower nectar (tting of
vectors or factor onto an ordination: r
2
0.218, P0.899). We
did not nd a monotonic relationship between height and
bacterial Shannon diversity index (Spearman's correlation
coecient: ρ=0.248, P= 0.254), but, in line with H3, a
unimodal distribution of bacterial Shannon diversity along the
height gradient with highest diversity in midelevation
(q(height, diversity) = 0.420, P= 0.011, Figure 6B).
FIGURE 1 Resource abundance and nectar attributes for owers of Marcgravia longifolia across strata. (A) Number of inorescences, (B) mean nectar
volume, (C) mean sugar concentration (% Brix), (D) mean sugar volume, and (E) mean pH of nectar for owers in understory (Low), midstory (Middle),
and canopy (High). Dierent letters indicate signicant dierences among strata. Shown are the predicted means with their 95% condence intervals derived
from the linear mixed eect models. Dots are individual raw data for (A) M. longifolia individuals or (BE) inorescences.
10 of 18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
DISCUSSION
By oering owers across a vertical gradient, M. longifolia
attracted a broad spectrum of dierent nectarivorous species.
Higher visitation of nectarivores in the underand midstory
were highly correlated with inorescence abundance. How-
ever, the diering inorescence abundance among strata was
not the only factor inuencing the dierences in the
interaction frequency of hummingbirds among strata. Even
though the hummingbird community composition did not
dier among strata, the dierences between subfamilies were
signicant, with hermits mainly foraging in the lower strata,
whereas trochilines foraged across the whole vertical gradient.
Further, the diversity of nectarinhabiting bacteria community
signicantly diered among strata with the highest Shannon
diversity index in the midstory. Our results illustrate the
importance of dierentiating among forest strata when
analyzing plantnectarivore interactions and highlight that
more than one determinant drives their vertical stratication.
Our ndings indicate that, unlike frugivores that have
strong preferences for a certain vertical foraging niche
(Thiel et al., 2023), nectarivores are more exible in their
use of vertical strata. Resource quantity and quality indeed
were decisive for inuencing the vertical foraging niche.
They have a stronger inuence on the foraging behavior of
nectarivores than on frugivores, and the presence of an
attractive resource across the whole vertical gradient drives
them to forage across strata. Resource availability is
considered as one of the most important drivers of network
structure, especially in communities of nectarivorous bats
and hummingbirds (Vázquez, et al., 2009a,2009b). These
hummingbirds derive around 90% of their dietary require-
ments from oral nectar (Gass and Montgomerie, 1981)
and, due to their extremely high metabolic rates imposed by
small size and hovering ight (Suarez and Welch, 2017),
need to visit hundreds of owers per day (Hurly, 1996).
Thus, for the small nectarivores, optimal foraging plays a
crucial role; for instance, they prefer to forage where many
rewarding owers are spatially close, and they tend to
proceed to the nearest, largest ower (Pyke, 1981). The
presence of a nearby attractive resource likely drives
nectarivorous species to leave their preferred vertical
foraging niche and to forage across all vertical strata. Most
observed hummingbird species visited M. longifolia owers
FIGURE 2 Nectar attributes for owers of Marcgravia longifolia over time. (A) Mean nectar volume. (B) Mean sugar concentration during anthesis. (C)
Relative nectar volume and (D) relative sugar concentration over 24 h. In A and B, dierent letters indicate signicant dierences among strata. Shown are
the predicted means with 95% condence intervals derived from the linear mixed eect models. Dots are individual raw data points for inorescences. In C
and D, black arrows represent the mean angle; bars represent the relative intensity of mean nectar production or mean sugar concentration for each time.
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
11 of 18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIGURE 3 Bipartite diagram depicting the interaction matrix of hummingbirds with Marcgravia longifolia in a tropical forest in northeastern Peruvian
Amazonia.Intotal,weobservedninetrochilineandthreehermithummingbirdspeciesvisitinginorescences in the three strata of M. longifolia.Thethickness
of the grey lines connecting hummingbird species and strata corresponds to the interaction frequency with which hummingbirds fed in the respective stratum.
12 of 18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
across strata. Even species previously described as typical
understory foragers, such as H. aurescens, P. philippii, and
P. superciliosus (Schulenberg et al., 2010), occasionally
moved up to forage in the canopy. Nevertheless, humming-
bird visitation decreased toward the canopy, most likely due
to the lower resource abundance. Although nectar volume
and sugar concentration did not dier among strata,
the abundance of inorescences was signicantly lower in
the canopy. Further, we assume that ying up and down the
vertical gradient comes with a high energy cost for small
FIGURE 4 Foraging behavior of hermit and trochiline hummingbirds across strata. (A) Hummingbird visitation, (B) Pilou's evenness index of hermits
and trochilines among strata. Dierent letters indicate signicant dierences among strata. Shown are the predicted means with their 95% condence
intervals derived from the linear, or generalized linear mixed eect models, respectively. Dots are individual raw data points for (A) M. longifolia individuals
or (B) hummingbird species, respectively.
FIGURE 5 Visitation to owers of Marcgravia longifolia by nectarivores. (A) Total number of visitations, (B) number of visitations by dierent
nocturnal and diurnal species (focal visitation) shown for the understory (Low), midstory (Middle), and canopy (High), (C) nocturnal visitation rate
shown for before, during, and after anthesis, and (D) nocturnal visitation rate in relation to sugar concentration. Dierent letters (ag) indicate signicant
dierences among strata. Shown are the predicted means with 95% condence intervals derived from the linear or generalized linear mixed eect models.
Dots are raw data points for M. longifolia individuals in A and B and for inorescences in C and D. The scatterplot shows tted data (circles), eect size
(solid line) and 95% condence intervals (dashed lines) of the linear mixed eects model.
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
13 of 18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
hummingbirds. Moreover, the microenvironmental condi-
tions in the canopy (Fetcher et al., 1985; Shaw, 2004) are
considered less favorable for their energy budget (Shankar
et al., 2019). In our system, the maximum temperature and
light intensity during the day were higher in the canopy
than at lower strata, whereas canopy cover was signicantly
lower. The more extreme microenvironmental conditions in
the canopy did not seem to inuence nectar attributes
across strata because nectar quantity and quality did not
signicantly dier among strata, even though they were
collected when conditions were more extreme. Further,
because hummingbirds have very good spatial memory
capabilities (Hurly, 1996), they likely remember the location
of rewarding owers and that the low abundance of
inorescences in the canopy renders it a nonor less
rewarding foraging location. The visitation of nectarivorous
bats was higher than that of hummingbirds across all strata.
This higher bat visitation across strata is crucial for
M. longifolia; nocturnal nectarivores are their main
pollinators, whereas hummingbirds act only as nectar
thieves. According to previous studies (Kalko and
Handley, 2001; Fleming et al., 2005), we assume that the
three identied nectarivorous bat species H. thomasi, C.
minor, and A. caudifer also foraged across all strata of M.
longifolia. Generally, bats are rather exible and frequently
use all forest strata (Kalko and Handley, 2001). In our
system, their visitation was highest in the midstory and thus
not as concurring with inorescence abundance as the
visitation of hummingbirds. Nectar production of M.
longifolia has a circadian rhythm, and the higher nectar
volume and sugar concentration during the night rendered
bats less dependent on the mere number of inorescences.
Even though temporal data on nectar could only be
collected for inorescences in the understory, we assume
this data to be representative because nectar attributes did
not even dier during the day when microenvironmental
dierences among strata were highest. Furthermore, it
might be easier for bats to approach inorescences in the
midstory where the surrounding vegetation is less dense
(Kalko and Handley, 2001; Diniz et al., 2019). Even in dry
forests where the vegetation is primarily low shrubs, bats
prefer the higher layers of the vegetation, whereas
hummingbirds are also found in the ground layer (Martins
and Batalha, 2007; Gottsberger and Silberbauer
Gottsberger, 2018; DomingosMelo et al., 2023). Consider-
ing the importance of bats as pollinators for M. longifolia,
collecting higherresolution data such as identifying indi-
vidual species and analyzing their community composition
across strata would be highly interesting.
Even though hummingbirds foraged across strata on an
attractive resource that was available along the entire vertical
gradient, the previously observed structure of neotropical
hummingbird communities was not dissolved, and dierent
species preferred a certain vertical niche for foraging.
Marcgravia longifolia attracted both hermit and trochiline
species and, in line with previous studies (Fleming et al., 2005;
Fleming and Kress, 2013), we still found the basic dichotomy
between hermit and nonhermit species. The three observed
hermit species mainly foraged in the underand midstory, only
rarely moving into the canopy. In contrast, the trochilines
frequently foraged across the whole vertical gradient. Competi-
tion for food resources has been described as one of the primary
drivers determining community organization in hummingbirds
(Feinsinger and Colwell, 1978;Wolf,1978; Ornelas et al., 2007).
Thus, especially in tropical lowland forests, high degrees of
specialization on specicoral resources are crucial for the
coexistence of the highly diverse hummingbird species
(Maglianesi et al., 2015). Our results suggest that niche division
among tropical hummingbirds might not only be dened by
specialization for certain owering plant species, but that
hummingbirds also divide the vertical gradient of a plant that is
attractive to both hermit and nonhermit species into distinct
vertical foraging niches. These observations are in line with the
studybyNaikatinietal.(2022) reporting that the vertical
foraging height of honeyeaters was formed by interspecic
competition. A further driver of niche division among hermit
FIGURE 6 Characteristics of bacterial communities in the nectar of M. longifolia across strata. (A) Composition across dierent heights, derived from
the NMDS model. Blue: low heights, red: high heights. (B) Shannon diversity index for bacterial communities in nectar from owers at dierent heights.
14 of 18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and trochiline species is their diering feeding behavior.
Whereas territorial trochilines establish and vigorously defend
territories (Feinsinger and Colwell, 1978), traplining hermits
travel among clumps of owers, probably following a regular
route and visiting these clumps of owers in a particular
sequence (Stiles, 1975). We also observed that trochiline species
established and defended territories around M. longifolia
individuals. They frequently chased away other hummingbirds,
their own and other species, which gave them exclusive access
to inorescences across all vertical strata. Hermits, on the other
hand, visited M. longifolia individuals quickly, always feeding on
the same under,andmidstoryinorescences in a regular,
sequential fashion. This behavior highlights how important
niche dierentiation, not only among plant species but also
among strata, is for reducing competition in tropical humming-
bird communities. Thus, the vertical niche can be a strong
structuring factor in plantanimal interactions and needs to be
further studied.
Because nectarinhabiting bacteria can be important
inuencers of plantanimal interactions (Vannette et al., 2013),
studying their communities could reveal important processes
driving the structure of plantanimal networks. Crucially, we
found that the diversity but not the composition of nectar
inhabiting bacteria communities of M. longifolia diered along
the vertical gradient with the highest Shannon diversity index
in the midstory. While this somewhat unclear pattern might
partially be attributed to the rather small number of samples,
we here discuss potential mechanisms that could lead to a
pattern of higher diversity in the midstory. We assume that the
many dierent nectarivorous species that frequently forage for
nectar of M. longifolia and move throughout the vertical
gradient had transferred bacteria to the nectar across strata,
with the highest likelihood of transfer to the midstory, which
they visited most frequently. Nectarivores are known to act as
vectors and transport microorganisms between owers
(Lachance et al., 2001b;Belisleetal.,2012;Sharabyetal.,2020).
Due to dierences in foraging patterns and morphology,
dierent nectarivores can transfer dierent microorganisms to
dierent spaces (Sharaby et al., 2020). For instance, Belisle et al.
(2012) attributed the nonrandom distribution of microfungi
inhabiting the nectar of Mimulus aurantiacus to spatially
nonrandom foraging by pollinators. In our study, the hermit
species supposedly followed a regular route from one
M. longifolia individual to the next, likely transferring bacteria
among individuals and thereby contributing to the high
bacterial diversity in the midstory. The trochiline species, on
the other hand, frequently foraged across strata and likely
transferred bacteria within M. longifolia individuals. Further,
the higher temperatures and light intensities in the lessdense
canopy may also have contributed to the lower bacterial
diversity in this stratum. In keeping with the potential of
nectarinhabiting bacteria as hidden playersin
plantnectarivore interactions (Vannette et al., 2013), the
vertical stratication of the nectarinhabiting bacterial commu-
nities thus likely inuences dierences among strata in the
patterning of plantnectarivore interactions.
CONCLUSIONS
We found dierences among strata in the patterning of
plantnectarivore interactions within a single plant species.
Our results suggest that vertical stratication in
plantnectarivore interactions is driven by resource abun-
dance, but that other factors such as speciesspecic
preferences of nectarivores for certain strata due to competi-
tion might play an important role. Further, the diversity of
nectarinhabiting bacteria communities was vertically strati-
ed. Considering the understudied inuence of bacteria on
plantnectarivore interactions (Vannette et al., 2013), their
vertically stratied structure might further drive dierences
in plantnectarivore interactions among strata. It is a long
held tenet in ecology that tropical species are very specialized
and have very negrained niche partitioning, which may
facilitate the coexistence of this high number of species in
tropics (Schemske, 2002). Niche dierentiation of species
along a vertical gradient may thus be a key factor promoting
diversity in tropical forests.
AUTHOR CONTRIBUTIONS
S.T., M.G., E.W.H., M.T., R.J., and K.H. conceived the ideas
and designed the methodology. S.T., M.G., J.K., N.L., and
N.S. collected the data. S.T., M.G., J.K., N.L., R.J., and K.H.
analyzed and interpreted the data. S.T., E.W.H., R.J., and
K.H. led the writing of the manuscript. All authors
contributed critically to the drafts and gave nal approval
for publication.
ACKNOWLEDGMENTS
We thank the German Research Foundation (Deutsche
Forschungsgemeinschaft, DFG) for funding the project
Vertical stratication of plantanimal interactions and their
impact on pollination and seed dispersal within a single
Neotropical plant speciesto K.H. (HE 7345/51), E.W.H. (HE
1870/271), and M.T. (TS 81/141). We further thank the Eva
MayrStihl foundation for their support. We are especially
grateful to Madita Jappe and Luca Hahn for their support
during the eld work and to Milagros Rimachi for identifying
the recorded plant species, to Robert S. Voss (American
Museum of Natural History) for identifying marsupials, Axel
Hausmann and Hubert Thony (Zoologische Staatssammlung
München) for identifying moths from cameratrap photos,
Alessandro Mainardi for advice with the circular statistics, and
to the reviewers for their constructive comments. We are
grateful to the Servicio Forestal y de Fauna Silvestre (SERFOR)
of the Peruvian Ministry of Agriculture for issuing research
permits (nos. 3042018MINAGRISERFORDGGSPFFS and
5282019MINAGRISERFORDGGSPFFS). Open Access
funding enabled and organized by Projekt DEAL.
DATA AVAILABILITY STATEMENT
Data of this paper are available from https://zenodo.org/
records/10527420 and https://qiita.ucsd.edu/study/description/
15418 (accession PRJEB72152 ERP156935).
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
|
15 of 18
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
ORCID
Sarina Thiel http://orcid.org/0000-0002-3652-9083
Malika Gottstein http://orcid.org/0000-0003-0359-7730
Eckhard W. Heymann http://orcid.org/0000-0002-
4259-8018
Marco Tschapka http://orcid.org/0000-0001-9511-6775
Robert R. Junker http://orcid.org/0000-0002-7919-9678
Katrin Heer http://orcid.org/0000-0002-1036-599X
REFERENCES
Allee, W. C., O. Park, A. E. Emerson, T. Park, and K. P. Schmidt. 1949.
Principles of animal ecology. WB Saunders, Philadelphia,
PA, USA.
ÁlvarezPérez, S., C. M. Herrera, and C. de Vega. 2012. Zoomingin on
oral nectar: a rst exploration of nectarassociated bacteria in wild
plant communities. FEMS Microbiology Ecology 80: 591602.
Amorim, F. W., C. S. Ballarin, G. Spicacci, G. Bergamasco, L. Carvalho,
W. Uieda, and A. P. Moraes. 2023. Opossums and birds facilitate the
unexpected bat visitation to the groundowering Scybalium
fungiforme.Ecology 104: e3935.
Angly,F.E.,P.G.Dennis,A.Skarshewski, I, Vanwonterghem, P. Hugenholtz,
and G. W. Tyson. 2014. CopyRighter: a rapid tool for improving the
accuracy of microbial community proles through lineagespecicgene
copy number correction. Microbiome 2: 11.
Basset, Y., P. M. Hammond, H. Barrios, and J. D. Holloway. 2003. Vertical
stratication of arthropod assemblages. In Y. Basset, R. Kitching, S.
Miller, and V. Novotný [eds.], Arthropods of tropical forests, 1727.
Cambridge University Press, Cambridge, UK.
Belisle, M., K. G. Peay, and T. Fukami. 2012. Flowers as islands: spatial
distribution of nectarinhabiting microfungi among plants of
Mimulus aurantiacus, a hummingbirdpollinated shrub. Microbial
Ecology 63: 711718.
Blüthgen, N., F. Menzel, T. Hovestadt, B. Fiala, and N. Blüthgen. 2007.
Specialization, constraints, and conicting interests in mutualistic
networks. Current Biology 17: 341346.
Bogo, G., A. Fisogni, J. RabassaJuvanteny, L. Bortolotti, M. Nepi,
M. Guarnieri, L. Conte, and M. Galloni. 2021. Nectar chemistry is
not only a plant's aair: oral visitors aect nectar sugar and amino
acid composition. Oikos 130: 11801192.
Bongers, F. 2001. Methods to assess tropical rain forest canopy structure:
an overview. Plant Ecology 153: 263277.
Brooks, M., B. Bolker, K. Kristensen, M. Maechler, A. Magnusson,
M. McGillycuddy, H. Skaug, et al. 2020. Tools for generalized linear
mixed models using template model builder. Website: https://cran.r-
project.org/web/packages/glmmTMB/index.html
BryschHerzberg, M. 2004. Ecology of yeasts in plantbumblebee mutual-
ism in Central Europe. FEMS Microbiology Ecology 50: 87100.
BuchananSmith, H. M., S. M. Hardie., C. Caceres, and M. J. Prescott. 2000.
Distribution and forest utilization of Saguinus and other primates of
the Pando Department, northern Bolivia. International Journal of
Primatology 21: 353379.
Canto, A., and C. M. Herrera. 2012. Microorganisms behind the
pollination scenes: microbial imprint on oral nectar sugar variation
in a tropical plant community. Annals of Botany 110: 11731183.
Carnicer,J.,P.Jordano,andC.J.Melián.2009.Thetemporaldynamicsof
resource use by frugivorous birds: a network approach. Ecology 90:
19581970.
Carter, C., S. Shar, L. Yehonatan, R. G. Palmer, and R. Thronburg. 2006.
A novel role for proline in plant oral nectars. Naturwissenschaften
93: 7279.
Chmel, K., J. Riegert, L. Paul, and V. Novotný. 2016. Vertical stratication
of an avian community in New Guinean tropical rainforest.
Population Ecology 58: 535547.
Czenze, Z.J., and T., Thurley. 2021. Dactylanthus ower visitation by New
Zealand lesser shorttailed bats appears to be inuences by daily
rainfall. New Zealand Journal of Ecology 45: 3436.
de Vega, C., C. M. Herrera, and S. D. Johnson. 2009. Yeasts in oral nectar
of some South African plants: quantication and associations with
pollinator type and sugar concentration. South African Journal of
Botany 75: 798806.
Diniz, U. M., A. DomingosMelo, and I. C. Machado. 2019. Flowers up!
The eect of oral height along the shoot axis on the tness of bat
pollinated species. Annals of Botany 124: 809818.
DomingosMelo, A., S. AlbuquerqueLima, U. M. Diniz, A. V. Lopes, and
I. C. Machado. 2023. Bat pollination in the Caatinga: a review of
studies and peculiarities of the system in the New World's largest and
most diverse Seasonally Dry Tropical Forest. Flora 305: 152332.
DomingosMelo, A., A. A. Cocucci, M. Tschapka, and I. C. Machado. 2023.
A negative association between nectar standing crop and pollen
transfer suggests nectar functions as a manipulator of pollinating
bats. Annals of Botany 131: 361372.
Dormann, C. F. 2011. How to be a specialist? Quantifying specialisation in
pollination networks. Network Biology 1: 120.
Dormann, C. F., J. Frund, N. Blüthgen, and B. Gruber. 2009. Indices,
graphs and null models: analyzing bipartite ecological networks. Open
Journal of Ecology 2: 724.
Encarnación, F. 1985. Introducción a la ora y vegetación de la Amazonia
peruana: estado actual de los estudios, medio natural y ensayo de una
clave de determinación de las formaciones vegetales en la llanura
amazónica. Candollea 40: 237252.
Feinsinger, P., and R. K. Colwell. 1978. Community organization among
Neotropical nectarfeeding birds. American Zoologist 18: 779795.
Fetcher, N., S. F. Oberbauer, and B. R. Strain. 1985. Vegetation eects on
microclimate in a lowland tropical forest in Costa Rica. International
Journal of Biometeorology 29: 145155.
Fleming, T. H.. 1993. Plantvisiting bats. American Scientist 81: 460.
Fleming, T. H., and W. J. Kress. 2013. The ornaments of life: coevolution
and conservation in the tropics. University of Chicago Press, Chicago,
IL, USA.
Fleming, T. H., and N. Muchhala. 2008. Nectarfeeding bird and bat niches
in two worlds: pantropical comparisons of vertebrate pollination
systems. Journal of Biogeography 35: 764780.
Fleming, T. H., N. Muchhala., and J. F. Ornelas. 2005. New World nectar
feeding vertebrates: community patterns and processes. In V.
SánchezCordero and R. A. Medellín [eds.], Contribuciones masto-
zoológicas en homenaje a Bernardo Villa, 163186.; Instituto de
Biología and Instituto de Ecología, Universidad Nacional Autonóma
de México (UNAM); Comisión Nacional para el Conocimiento y Uso
de la Biodiversidad (CONABIO); Mexico City, Mexico.
Fox, J., and S. Weisberg. 2019. An R companion to apllied regression.
Website: https://cran.r-project.org/web/packages/car/citation.html
Fridman, S., I. Izhaki, Y. Gerchman, and M. Halpern. 2012. Bacterial
communities in oral nectar. Environmental Microbiology Reports 4:
97104.
Gass,C.L.,andR.D.Montgomerie.1981.Hummingbirdforagingbehaviour:
decisionmaking and energy regulation. In A. Kamil and T. Sargnet [eds.],
Foraging behavior: ecological, ethological and physiological approaches,
159196. Garland STPM Press, NY, NY, USA.
Gaube, P., R. R. Junker, and A. Keller. 2021. Changes amid constancy:
ower and leaf microbiomes along land use gradients and between
bioregions. Basic and Applied Ecology 50: 115.
Golonka, A. M., B. Obi Johnson, J. Freeman, and D. J. Hinson. 2014.
Impact of nectarivorous yeasts on Silene caroliniana's scent. Eastern
Journal of Agricutural and Biologial Sciences 3: 127.
Gottsberger,G.,andI.SilberbauerGottsberger. 2018. How are pollination and
seed dispersal modes in Cerrado related to stratication? Trends in a
cerrado sensu stricto woodland in southeastern Brazil, and a comparison
with Neotropical forests. Acta Botanica Brasilica 32: 434445.
GonzálezCastro, A., S. Yang, M. Nogales, and T. A. Carlo. 2012. What
determines the temporal changes of species degree and strength in an
oceanic island plantdisperser network? PLoS one 7: e41385.
Griessenberger, F., W. Trutschnig, and R.R. Junker. 2022. qad:An
Rpackage to detect asymmetric and directed dependence in bivariate
samples. Methods in Ecology and Evolution 13: 21382149.
16 of 18
|
VERTICAL STRATIFICATION IN A LIANA FLOWERING ACROSS STRATA
15372197, 2024, 3, Downloaded from https://bsapubs.onlinelibrary.wiley.com/doi/10.1002/ajb2.16303, Wiley Online Library on [01/05/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Helletsgruber, C., S. Dötterl, U. Ruprecht, and R. R. Junker. 2017. Epiphytic
bacteria alter oral scent emissions. Journal of Chemical Ecology 43:
10731077.
Herrera, C. M., C. de Vega, A. Canto, and M. I. Pozo. 2009. Yeasts in oral
nectar: a quantitative survey. Annals of Botany 103: 14151423.
Herrera, C. M., I. M. García, and R. Pérez. 2008. Invisible oral larcenies:
microbial communities degrade oral nectar of bumble bee
pollinated Plants. Ecology 89: 23692376.
Herrera, C. M., and M. I. Pozo. 2010. Nectar yeasts warm the owers of a
winterblooming plant. Proceedings of the Royal Society, B, Biological
Sciences 277:18271834.
Heymann, E. W., S. Dolotovskaya, and E. R. Tirado Herrera. 2021. Estación
Biológica Quebrada Blanco. Biotropica 23: 202101.
Heymann, E. W., and E. R. Tirado Herrera. 2021. Estación Biológica
Quebrada Blanco un sitio poco conocido para investigación en
biodiversidad y ecología en la Amazonía peruana. Revista Peruana de
Biología 28: e2026.
Howe, F., and J. Smallwood. 1982. Ecology of seed dispersal. Annual
Review of Ecology, Evolution, and Systematics 13: 201228.
Hurly, T. A. 1996. Spatial memory in rufous hummingbirds: memory for
rewarded and nonrewarded sites. Animal Behaviour 51: 177183.
Jordano, P. 2000. Fruits and frugivory. In Fenner M [ed], The ecology of
regeneration in plant communities, 125165. CAB International,
Wallingford, UK.
Junker, R. R., F. Griessenberger, and W. Trutschnig. 2021. Estimating
scaleinvariant directed dependence of bivariate distributions.
Computational Statistics & Data Analysis 153: 107058.
Junker, R. R., M. Hanusch, X. He, V. RuizHernández, J. C. Otto,
S. Kraushaar, K. Bauch, et al. 2020. Ödenwinkel: an alpine platform
for observational and experimental research on the emergence of
multidiversity and ecosystem complexity. Web Ecology 20: 95106.
Junker, R. R., and A. Keller. 2015. Microhabitat heterogeneity across leaves
and ower organs promotes bacterial diversity. FEMS Microbiology
Ecology 91: 97.
Junker, R. R., T. Romeike, A. Keller, and D. Langen. 2014. Density
dependent negative responses by bumblebees to bacteria isolated
from owers. Apidologie 45: 467477.
Kalko, E. K. V., and C. O. J. Handley. 2001. Neotropical bats in the canopy:
diversity, community structure, and implications for conservation.
Plant Ecology 153: 319333.
Klingbeil, B. T., and M. R. Willig. 2008. Guildspecic responses of bats to
landscape composition and conguration in fragmented Amazonian
rainforest. Journal of Applied Ecology 46: 203213.
Kobayashi, S., T. Denda, C. C. Liao, Y. H. Lin, J. Placksanoi,
S. Waengsothorn, C. Aryuthaka, et al. 2020. Eects of dierent
pollinators and herbivores on the fruit set height of the mammal
pollinated treeclimbing vine Mucuna macrocarpa.Silviculture and
Plant Sciences 25: 315321.
Lachance, M. A., J. M. Bowles, S. Kwon, G. Marinoni, W. T. Starmer, and
D. H. Janzen. 2001a. Metschnikowia lochheadii and Metschnikowia
drosophilae, two new yeast species isolated from insects associated
with owers. Canadian Journal of Microbiology 47: 103109.
Lachance, M. A., W. T. Starmer, C. A. Rosa, J. M. Bowles, J. S. F. Barker,
and D. H. Janzen. 2001b. Biogeography of the yeasts of ephemeral
owers and their insects. FEMS Yeast Research 1: 18.
Lenth, R., B. Bolker, P. Buerkner, I. GinéVásquez, M. Herve, M. Jung,
J. Love, et al. 2020. emmeans: Tools for estimated marginal means,
aka leastsquares means. R package version 1.4.8. Website: https://
cran.r-project.org/web/packages/emmeans/index.html
Lievens, B., J. E. Hallsworth, M. I. Pozo, Z. B. Belgacem, A. Stevenson,
K. A. Willems, and H. Jacquemyn. 2015. Microbiology of sugarrich
environments: diversity, ecology and system constraints. Environmental
Microbiology 17: 278298.
Lobova, T. A., C. K. Geiselman, and S. A. Mori. 2009. Seed dispersal by bats
in the neotropics. New York Botanical Garden Press, Bronx, NY, USA.
e, T. M., E. R. Tirado Herrera, M. Nadjafzadeh, P. Berles, A. C. Smith,
C. Knogge, and E. W. Heymann. 2018. Seasonal variation and an
outbreakof frog predation by tamarins. Primates 59: 549552.
Maglianesi, M. A., N. Blüthgen, K. BöhningGaese, and M. Schleuning.
2015. Functional structure and specialization in three tropical
planthummingbird interaction networks across an elevational
gradient in Costa Rica. Ecography 38: 11191128.
Martins, F. Q., and M. A. Batalha. 2007. Vertical and horizontal
distribution of pollination systems in cerrado fragments of Central
Brazil. Brazilian Archives of Biology and Technology 50: 503514.
McCaig, T., L. Sam, A. Nakamura, and N. E. Stork. 2020. Is insect vertical
distribution in rainforests better explained by distance from the canopy
top or distance from the ground? Biodiversity and Conservation 29:
10811103.
Naikatini, A. N., G. Keppel, G. Brodie, and S. Kleindorfer. 2022.
Interspecic competition and vertical niche partitioning in Fiji's
forest birds. Diversity 14: 223.
Oksanen, J., G. L. Simpson, F. G. Blanchet, R. Kindt, P. Legendre,
P. R. Minchin, R. B. O'Hara, et al. 2019. egan: Community ecology
package. Website: https://cran.r-project.org/web/packages/vegan/
index.html
Ornelas, J. F., M. Ordano, and C. Lara. 2007. Nectar removal eects on seed
production in Moussonia deppeana (Gesneriaceae), hummingbird
pollinated shrub. Ecoscience 14: 117123.
Paulson, J. N., N. D. Olson, D. J. Braccia, J. Wagner, H. Talukder, M. Opo,
and H. C. Bravo. 2013. Dierential abundance analysis for microbial
marker