Repeated evolution of pollination syndromes in a highly diverse bromeliad lineage is correlated with shifts in life form and habitat

Abstract

Mutualistic plant-pollinator interactions play a critical role in the diversification of flowering plants. The spatiotemporal correlates of such interactions can be understood in a phylogenetic context. Here, we generate ddRAD-seq data for the highly diverse Vriesea-Stigmatodon lineage to test for correlated trait evolution among pollination syndromes and life form, habitat type, and altitude. Our results show that pollination syndromes are correlated with changes in life form and habitat type. The ancestor of the Vriesea-Stigmatodon lineage was likely bat pollinated, rock dwelling and inhabited open, mid-elevation forests. Transitions from bat to hummingbird pollination are correlated with transitions to the epiphytic life form in shaded habitats, whereas bat pollination is correlated with the rock-dwelling life form and open habitats. Our dated phylogenetic tree reveals independent origins of hummingbird pollination, occurring twice in Vriesea at c. 5.8 and 5.4 Mya. The timing for the shifts in pollination syndrome coincides with geological and environmental transformations across the Serra do Mar Mountain Chain, which increased habitat heterogeneity where Vriesea and their mutualists diversified. The phylogenetic tree reinforces the non-monophyly of taxonomic sections within the genus Vriesea previously defined by flower morphology, indicating that some lineages should be treated as species complexes. This study identifies synergetic drivers of speciation in a tropical biodiversity hotspot.

Full text

© 2023 e Linnean Society of London.
is is an Open Access article distributed under the terms of the Creative Commons Aribution-NonCommercial License (hps://creativecommons.org/licenses/by-nc/4.0/),
which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.
permissions@oup.com
Received 3 November 2022; revised 16 January 2023; accepted 12 May 2023
Original Article
Repeated evolution of pollination syndromes in a highly
diverse bromeliad lineage is correlated with shis in life form
and habitat
BeatrizNeves1,2,3,*,ǂ,, Paola de L.Ferreira3,4,6,*,ǂ, FranciscoProsdocimi5, Igor M.Kessous1,2,3,,
Dayvid R.Couto1, Ricardo L.Moura1, FabianoSalgueiro7,, Andrea F.Costa1,,
Christine D.Bacon2,3,§, AlexandreAntonelli,2,3,8,9,
1Museu Nacional, Universidade Federal do Rio de Janeiro, Quinta da Boa Vista, São Cristóvão, Rio de Janeiro, RJ, 20940-040, Brazil
2Department of Biological and Environmental Sciences, University of Gothenburg, Carl Skosbergs Gata 22B, Göteborg, SE 41319, Sweden
3Gothenburg Global Biodiversity Centre, Carl Skosbergs Gata 22B, Göteborg, SE 41319, Sweden
4Department of Biology, Aarhus University, Ny Munkegade 116, 8000 Aarhus C, Denmark
5Laboratório de Genômica e Biodiversidade, Instituto de Bioquímica Médica Leopoldo de Meis, Universidade Federal do Rio de Janeiro, Rio de
Janeiro, RJ, 21941-902, Brazil
6Departamento de Biologia, Faculdade de Filosoa Ciências e Letras de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, SP, 14051-901,
Brazil
7Departamento de Botânica, Universidade Federal do Estado do Rio de Janeiro, Av. Pasteur 458, Rio de Janeiro, RJ, 22290-240, Brazil
8Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3AE, UK
9Department of Plant Sciences, University of Oxford, South Parks Road, Oxford, OX1 3RB, UK
ǂShared rst authorship
§Shared last authorship
*Corresponding authors: Museu Nacional, Universidade Federal do Rio de Janeiro, Quinta da Boa Vista, São Cristóvão, Rio de Janeiro, 20940-040, Brazil. Department of
Biology, Aarhus University, Ny Munkegade 116, 8000 Aarhus C, Denmark. E-mail: beatriznevesbio@gmail.com; paolaferreira@bio.au.dk
ABSTRACT
Mutualistic plant-pollinator interactions play a critical role in the diversication of owering plants. e spatiotemporal correlates of such inter-
actions can be understood in a phylogenetic context. Here, we generate ddD-seq data for the highly diverse Vriesea-Stigmatodon lineage to test
for correlated trait evolution among pollination syndromes and life form, habitat type, and altitude. Our results show that pollination syndromes
are correlated with changes in life form and habitat type. e ancestor of the Vriesea-Stigmatodon lineage was likely bat pollinated, rock dwelling
and inhabited open, mid-elevation forests. Transitions from bat to hummingbird pollination are correlated with transitions to the epiphytic life
form in shaded habitats, whereas bat pollination is correlated with the rock-dwelling life form and open habitats. Our dated phylogenetic tree
reveals independent origins of hummingbird pollination, occurring twice in Vriesea at c. 5.8 and 5.4 Mya. e timing for the shis in pollination
syndrome coincides with geological and environmental transformations across the Serra do Mar Mountain Chain, which increased habitat het-
erogeneity where Vriesea and their mutualists diversied. e phylogenetic tree reinforces the non-monophyly of taxonomic sections within the
genus Vriesea previously dened by ower morphology, indicating that some lineages should be treated as species complexes. is study identi-
es synergetic drivers of speciation in a tropical biodiversity hotspot.
Keywords: Atlantic Forest; bat pollination; ddD-seq; hummingbird pollination; Neotropics; phylogenomics
INTRODUCTION
Mutualistic plant-pollinator interactions play a critical role in
the diversication of owering plants, which represent c. 90%
of the extant plant species diversity on land (Fenster et al. 2004,
Crepet and Niklas 2009, WFO 2020). Such interactions have
been inuenced by abiotic factors (Blois et al. 2013, Condamine
et al. 2018), particularly in the tropics, where high and stable
temperature and precipitation levels are favourable to the
Botanical Journal of the Linnean Society, 2023, XX, 1–12
hps://doi.org/10.1093/botlinnean/boad015
Advance access publication 1 July 2023
Original Article
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

2 • Neves et al.
formation of diverse mutualistic interactions (Chomicki et al.
2019). Shis in pollination systems can promote plant species
diversication, increasing species diversity (van der Niet and
Johnson 2012, Givnish et al. 2014, Lagomarsino et al. 2016,
Serrano-Serrano et al. 2017). Yet, current knowledge on the
spatiotemporal evolution of plant-pollinator interactions and
their abiotic correlates remains fragmentary and is mostly
lacking for biologically complex ecosystems, such as tropical
rainforests.
e study of tropical species-rich plant clades, such as the
Neotropical plants bromeliads (Bromeliaceae), can improve our
understanding of biotic drivers of diversication. Among the
bromeliads, Vriesea Lindley is mostly restricted to the Atlantic
Forest and to the campos rupestres of the Cerrado savanna, two
biodiversity hotspots in Brazil (Myers et al. 2000, BFG 2018).
Vriesea includes c. 230 species and is especially diverse in the
Atlantic Forest, along the Serra do Mar Mountain Chain where
c. 85% of all species are endemic (BFG 2018, Gouda et al. con-
tinuously updated). Also, Stigmatodon Leme, G. K. Br. & Barfuss
was recently segregated from Vriesea as an independent genus
(Barfuss et al. 2016). Stigmatodon comprises 20 rupicolous spe-
cies endemic to the Brazilian inselbergs, occurring exclusively
on vertical granite rocky outcrops, with high exposure to solar
radiation (Barfuss et al. 2016, Couto et al. 2022, Gouda et al. con-
tinuously updated).
Most Vriesea species are epiphytes, with the rupicolous (rock
dwelling) and terrestrial life forms occurring less frequently
(BFG 2018). Vriesea occupy dierent forest strata where they
can form either large (more than 100 individuals) or small popu-
lations (Costa et al. 2014, BFG 2018). Vriesea interacts with two
main pollinator groups: hummingbirds and bats (Sazima et al.
1999, Buzato et al. 2000). Vriesea species pollinated by hum-
mingbirds have red to yellow oral bracts, tubular owers with
exserted stamens, no scent, and diurnal anthesis (Buzato et al.
2000). Conversely, the Vriesea owers pollinated by bats have
green to brown oral bracts and campanulate owers with in-
cluded stamens, are scented, and have nocturnal anthesis
(Sazima et al. 1999). e owers of Stigmatodon species are
adapted to bat pollination, being similar to the owers of Vriesea
with bat-pollination syndrome.
Pollination syndrome consists of a particular set of oral
traits, such as shape, colour, scent, and phenology (Faegri and
van der Pijl 1979). Syndromes are generally used to infer or pre-
dict unobserved pollinators (Fenster et al. 2004, Rosas-Guerrero
et al. 2014, Lagomarsino et al. 2017), including in Vriesea (Neves
et al. 2020a). However, this approach has been criticized for
oversimplifying complex plant-animal interactions (such as di-
urnal and nocturnal dierences in pollinators; Muchhala 2003)
that may lead to unreliable predictions (Ollerton et al. 2009,
Dellinger 2020).
Avian pollination, epiphytism, and the tank habit
(overlapping leaves that accumulate water and organic ma-
terial) are key innovations in bromeliad species (Givnish et al.
2014, Silvestro et al. 2014). Vriesea occur in habitat types asso-
ciated with high diversication rates in Bromeliaceae, such as
the tropical mountains on which the high diversity has arisen
from a combination of factors. Among them, (i) epiphytism
allows the plants to occupy a broader range of forest strata,
especially when associated with (ii) the tank habit, which, to-
gether with (iii) absorptive trichomes on the leaves, confers
independence from soil substrates (Givnish et al. 2014). In
addition, the association with (iv) dierent pollinator groups
capable of thermoregulation and ying over longer distances
(hummingbirds and bats), in contrast to the insects. Finally,
humid and steep tropical mountains have (v) abundant rainfall
availability throughout the year and (vi) a diversity of habitat
types isolated from each other that promotes speciation
(Givnish et al. 2014).
Dierences in both forest habitat types and ight paerns of
pollinator groups are used to explain the occurrence of plants
in either open or dense vegetation habitats. Bat-pollinated
plants oen occur exposed in open habitats to facilitate echo-
location and view by bats, though they can also occur in dense
vegetation when associated with ower scents that guide bats
(Muchhala and Serrano 2015). When hovering, bats sweep
their wings over a large area around their bodies, whereas hum-
mingbirds are more manoeuvrable, with wing movements re-
stricted to a subtle area directly behind their backs (Muchhala
2003).
Phylogenetic studies within Vriesea in the literature have
struggled to identify molecular markers capable to discriminate
infrageneric groups and delimit species boundaries (Costa et
al. 2015, Gomes-da-Silva and Souza-Chies 2017, Kessous et al.
2020, Machado et al. 2020, Loiseau et al. 2021). Using the phylo-
genetic tree from Gomes-da-Silva and Souza-Chies (2017);
Kessler et al. (2020) assessed the gain and loss of hummingbird
pollination in Vriesea to illustrate how hummingbirds may lead
to increased diversication rates. Kessler et al. (2020) identi-
ed three shis from hummingbird to bat pollination in Vriesea,
but they treated Stigmatodon (Barfuss et al. 2016) as Vriesea.
Furthermore, the study identied a gap in our understanding
of the correlates of distinct plant-pollinator interactions which
would allow for the identication of the drivers of diversication
of the clade.
In order to investigate diversication in the Vriesea-
Stigmatodon clade, we developed novel genomic plastid and
nuclear data generated with double digest restriction site as-
sociated DNA sequencing (ddD-seq; Peterson et al. 2012).
e use of D-seq resolved relationships within species
complexes and amongst closely related species with high reso-
lution (Eaton and Ree 2013, Leaché et al. 2014, Massai et al.
2016). In bromeliads, D-seq data of Alcantarea (É.Morren
ex Mez) Harms indicated that it is a powerful tool to investi-
gate genetic diversity between closely related species (Lexer
et al. 2016).
Here, we infer the evolution of plant-pollinator inter-
actions and their morphological and environmental correlates.
Specically, we test the correlated evolution of hummingbird
and bat pollination syndromes with life form, habitat type, and
altitude through time. We hypothesize that: the correlation with
hummingbirds is coupled to the epiphytic life form and occu-
pation of the shaded understory of the Atlantic Forest, whereas
bat pollination is associated with the rupicolous and terrestrial
life forms in open areas of the forest. Our study highlights the
intricate evolution of plant-pollinator interactions and identies
important drivers of tropical biodiversity.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

Evolution of pollination syndromes • 3
MATERIALS AND METHODS
Sampling
We sampled a total of 59 individual plants, including 47 acces-
sions of Vriesea and seven accessions of Stigmatodon, with good
representation of their morphological variation and geograph-
ical distribution (Supporting Information, Material S1). We
sampled around 20% of the hyperdiverse Vriesea (with a total
of 230 species, Gouda et al. continuously updated). Diculty
in extracting high-quality DNA, which is required for Next
Generation Sequencing techniques like ddD-seq, hampered a
more complete sampling. Vriesea was formerly treated as a single
genus (Smith and Doowns 1977), but was segregated into seven
genera distributed in two dierent subtribes (Cipuropsidinae
and Vrieseinae) consisting the tribe Vrieseeae (Grant 1995,
Barfuss et al. 2016, Leme et al. 2017). Species from the related
genera Alcantarea (subtribe Vrieseinae), Lutheria Barfuss &
W.Till, Goudaea W.Till & Barfuss (subtribe Cipuropsidinae), and
Tillandsia L. (tribe Tillandsieae, aer Barfuss et al. 2016) were
included as outgroups. We collected samples from natural popu-
lations mainly in the Atlantic Forest and Cerrado, prioritizing
type localities. Additional samples were collected from living
collections. Information on vouchers is presented in Supporting
Information, Material S1. Tillandsioideae species are mostly dip-
loid 2n = 50, as shown by previous studies (Cotias-de-Oliveira
et al. 2004, Palma-Silva et al. 2004, Ceita et al. 2008, Manhães
2021).
DNA extraction, library preparation, and sequencing
We extracted DNA from silica gel-dried leaf material and then
stored it at -80° C. Frozen leaf samples were crushed into powder
using a TissueLyser II (Qiagen). We extracted DNA using the
CTAB protocol of Doyle and Doyle (1987), with modica-
tions following Azmat et al. (2012). We determined DNA ex-
traction quality on a 1% agarose gel and quantied DNA using
a NanoDrop® spectrophotometer and a Qubit® uorimeter 3.0
(High sensitivity kit; Life Technologies).
We standardized DNA samples to concentrations of 10 ng/
µl and a total of 50 µl of each sample. Library preparation and
single-end DNA sequencing on an Illumina HiSeq 2000 was
performed by Floragenex Inc. (Eugene, OR, USA). Total DNA
was double-digested with the SbfI and PstI enzymes (ddD-
Seq, Peterson et al. 2012).
Genome assembly and mapping
We performed de novo assembly using the soware pipeline
PyD v.3.0.4 (Eaton 2014). Genomic data for each species
varied and PyD was used to assemble loci by optimizing
coverage across datasets. Optimization through an alignment-
clustering algorithm allowed for indel variation within and be-
tween samples, recovering more shared loci across disparate taxa
(Eaton 2014). We dened the following parameters: minimum
coverage per cluster = 6, clustering threshold = 0.85, minimum
sample coverage for loci = 40, maximum number of individuals
with shared heterozygous sites = 3, and the remaining param-
eters were set to default.
To identify the chloroplast loci in our dataset, we performed
a Bowtie search against the pineapple chloroplast genome
(GenBank accession number NC_026220.1), using Bowtie2
with the ‘--very-sensitive-local’ parameter (Langmead and
Salzberg 2012).
Phylogenetic analyses
Phylogenetic trees were inferred using Maximum Likelihood
(ML) and coalescent approaches. ML based on the concaten-
ated nuclear and plastid loci was inferred using xML-HPC2
on XSEDE via the CIPRES Science Gateway v.3.3 (Miller et
al. 2010), seing a GTR+GAMMA substitution model and
rapid bootstrap (BS) estimation based on 1000 replicates. We
interpreted BS values ≥ 90 as strong, 89–70 as moderate, and
69–50 as weak support (Hillis and Bull 1993). We also ran
ML using the PhyML online platform v.3.0 (hp://www.atgc-
montpellier.fr/phyml/) with the following parameters: auto-
matic model selection-AIC, tree searching nearest neighbour
interchange, and a branch support approximate likelihood-ratio
test Shimodaira–Hasegawa-like (aLRT SH), which is a simpler
and faster branch support test recommended for large molecular
datasets (Guindon et al. 2010).
Coalescent analyses were performed for the nuclear loci
using ASTL-III (Zhang et al. 2018) and SVDquartets
(Chifman and Kubatko 2014), following Ferreira et al. (2022).
Briey, ASTL-III was ran based on the unrooted trees es-
timated by ML searches in xML (Stamatakis 2014) and
branch support was evaluated using local posterior probabil-
ities (PP). SVDquartets was inferred with exhaustive sampling
all possible quartets. Branch support was evaluated using 1000
non-parametric BS replicates.
We estimated the divergence times using a penalized likeli-
hood approach in treePL (Smith and O’Meara 2012). We used
the xML tree with the concatenated dataset as input and
secondary calibrations at the Vrieseinae crown node (5.4–10.2
Mya) and at the Vriesea crown node (3.3–6.8 Mya) based on
Givnish et al. (2014) and Kessous et al. (2020). To calculate the
95% condence interval of node ages, we ran treePL on each
of the BS trees from xML, then used TreeAnnotator from
BEAST v.1.10.4 (Suchard et al. 2018) to generate a summary
tree. We visualized phylogenetic trees of all methods described
above using FigTree v.1.43 (Rambaut 2014).
Ancestral trait reconstruction
We estimated ancestral states for pollination syndrome
(hummingbird and bat), life form (epiphyte, terrestrial, and
rupicolous), habitat type (open and shade), and altitude (con-
tinuous values) to understand when and how many times these
traits evolved in Vriesea and Stigmatodon. Information on traits
was extracted from eld observations, monographs, oras, and
recent taxonomic reviews (Supporting Information, Material S2;
Smith and Downs 1977, Versieux and Wanderley 2008, Costa et
al. 2009, Moura, 2011, Nogueira 2013, Neves et al. 2018, 2020a,
Uribbe et al. 2020, Couto et al. 2022). We used the dated tree
from treePL built with the concatenated dataset for character
optimizations. For ancestral states reconstruction of discrete
traits, we used Bayesian stochastic character mapping (Bollback
2006) estimated from 1000 iterations with the function make.
simmap in the R package phytools (Revell 2012, R Core Team
2018). We coded pollination syndromes based on ower and
oral bract traits, following Neves et al. (2020a), who validated
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

4 • Neves et al.
the utility of pollination syndromes in Vriesea based on informa-
tion on conrmed pollinators from the literature. ere are some
Vriesea species known to be visited and/or pollinated by both
hummingbirds and bats (Sazima et al. 1995, Aguilar-Rodríguez
et al. 2019). ese studies show that bat-pollinated owers
can be visited by hummingbirds at dawn, but they feed on the
small amount of nectar le by the bats in withered owers. In
such cases, based on oral syndromes, we coded one group as
the primary pollinator, considering that pollinators ecacy and
eciency was not tested in these studies. For the continuous
trait of altitude, we used a ML reconstruction with the function
contMap from phytools. We used the mean altitude for each spe-
cies based on information extracted from the dataset compiled
in Ramos et al. (2019) for Atlantic Forest epiphytes, comple-
mented with data from our personal collections (Supporting
Information, Material S2).
Correlation among traits
In order to test the correlation of pollination syndrome with life
form, habitat type, and altitude, we ed Bayesian threshold
models (Felsenstein 2012, Revell 2012) using the function
threshBayes in phytools. We modelled pollination syndrome, life
form, and habitat type as discrete binary traits and altitude as a
continuous trait. We divided the multistate trait life form into
epiphyte or terrestrial/rupicolous. When coding life form, we
considered the predominant state for polymorphic species. In
addition, merging terrestrial and rupicolous states, in this spe-
cic case, brings greater statistical power to our analysis, as we
use fewer parameters. e great majority of Vriesea are epi-
phytes (c. 150 spp.) and the terrestrial and rupicolous Vriesea
are the minority (BFG 2018). We ran the Markov chain Monte
Carlo (MCMC) for 3 million generations sampling every 1000,
with a burn-in of 20% to summarize the posterior distribution
values for the correlation coecient (r). We calculated the ef-
fective sample size (ESS) of the coecient using the function
eectiveSize in the R package coda (Plummer et al. 2006).
Distribution paerns of hummingbird and bat pollination
syndromes
In order to infer distribution paerns of species from humming-
bird and bat pollination syndromes, we built spatial plots of
altitude vs. latitude. With this analysis we aim to detect if plant
species from hummingbird and bat pollination syndromes are
spatially segregated or not. Because distribution paerns are
likely correlated to preferences of habitat, physiology, and other
ecological aspects of their specic pollinator species (Aguilar-
Rodríguez et al. 2019, Kessler et al. 2020), we used data from
the literature on documented Vriesea pollinator interactions to
build a plant–pollinator network. Taken together, this approach
allowed a beer visualization and an integrated discussion of
Vriesea spatial distribution and pollinator-specic interactions.
ere is a complete lack of pollination studies for Stigmatodon,
so it was not included in this analysis. Further, the current 20
Stigmatodon species show a clear bat pollination syndrome and
their trait and spatial variation is captured within the variation in
Vriesea (Barfuss et al. 2016, Couto et al. 2022).
To map spatial distribution, we built plots using 9568 occur-
rence records for 132 species from the dataset of Ramos et al.
(2019). First, we generated boxplots for each species using alti-
tudinal data and excluded the outliers using the function boxplot
in the R package graphics (Murrell 2018). en, we used ggplot2
(Wickham 2011) to produce the altitude vs. latitude plots for
species of each syndrome.
In addition, we used the data on pollination biology compiled
by Neves et al. (2020a), including all reported Vriesea pollinator
interactions in both peer-reviewed and grey literature. We ex-
tracted information on the identity of plants and oral visitors
and pollinators. In total, we documented interactions between
35 Vriesea species and 16 hummingbird and three bat species
(Supporting Information, Material S3). e most representa-
tive hummingbird pollinators are Phaethornis eurynome Lesson,
Ramphodon naevius Dumont, alurania glaucopis Gmelin, and
Leucochloris albicollis Vieillot, interacting with 19, 12, 11, and
nine Vriesea species respectively. e bat species Anoura caudifer
É.Georoy is recorded visiting owers of each of the eight re-
gistered Vriesea species with bat pollination syndrome. We used
this information to interpret and discuss our results.
RE S ULTS
Genome assembly and mapping
We generated in total 85 GB of raw data containing 358 425 268
reads of 100 bp each. e nal dataset comprised 664 ddD-
seq loci, 11 of which were from the chloroplast genome. e
concatenated alignment totalled 57 640 bp in length, with 4878
variable sites, 1868 of which were parsimony informative sites.
e percentage of gaps and missing data was 21.99%.
Phylogenetic relationships and clade age
e ML tree resolved high support for most of the deep phylo-
genetic relationships (Fig. 1). Subtribe Vrieseinae (BS = 97,
aLRT SH = 1) and each genus were monophyletic, including
Stigmatodon (clade A, BS = 98, aLRT SH = 1) and Vriesea
(clade B, BS = 93, aLRT SH = 1). We identied two distinct
lineages of both hummingbird- and bat-pollinated species in
Vriesea, emerging from two main clades: C (BS = 87, aLRT
SH = 99) and D (BS = 54, aLRT SH = 28). Clades E (BS = 23,
aLRT SH = 99) and F (BS = 65, aLRT SH = 99) are two large
lineages of species exclusively from each of the two pollination
syndromes (Fig. 1). We recovered poor resolution for shallow
nodes; however, some groupings of morphologically similar
species gained high support. In other cases, dierent accessions
of the same species were not resolved as monophyletic, such as
Vriesea agostiniana E.Pereira. e topology shown here refers
to the ML tree inferred in xML, which presented few incon-
gruences in the shallow relationships within Vriesea when com-
pared with the ML tree inferred in PhyML and the coalescent
trees inferred with both Astral and SVDquartets (Supporting
Information, Materials S4, S5). ese incongruences are mostly
at poorly supported nodes.
We estimated the crown age of subtribe Vrieseinae in the Late
Miocene 10.1 million years ago (Mya) [95% High Posterior
Density (HPD): 10.18–10.19 Mya], Stigmatodon 8.0 Mya (95%
HPD: 7.4–8.6 Mya), and Vriesea 6.3 Mya (95% HPD: 5.5–6.7
Mya). All shis among syndromes in Vriesea occurred between
5.8 and 5.4 Mya (95% HPD: 4.4–6.2 Mya; Fig. 2).
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

Evolution of pollination syndromes • 5
Ancestral trait reconstruction and evolutionary correlates of
pollination syndrome
For Vriesea, we inferred a bat pollination syndrome as the an-
cestral state (PP = 0.95; Fig. 3A), and two independent shis to
hummingbird pollination. For habitat, the ancestral condition
was inferred to be shaded environments (PP = 0.99; Fig. 3B)
and we recovered seven independent transitions to open areas.
e epiphytic life form was ancestral (PP = 0.99; Fig. 3C), with
Figure 1. ML tree of Vriesea and Stigmatodon based on 664 ddD-seq loci showing hummingbird-pollinated and bat-pollinated lineages.
BS support values above 50% are shown at internodes. For selected nodes discussed in the text, we show approximate likelihood-ratio test
Shimodaira–Hasegawa-like (aLRT SH-like) support. Pollination syndromes are shown to evolve repeatedly in Vriesea.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

6 • Neves et al.
the terrestrial and rupicolous life forms evolving multiple times
in this hyperdiverse genus (at least 12 and 10 times, respect-
ively). For Stigmatodon, we recovered bat pollination syndrome
(PP = 1.00; Fig. 3A), open environments (PP = 0.94; Fig. 3B),
and rupicolous life form (PP = 0.97; Fig. 3C) as ancestral states.
For the larger clade Vriesea-Stigmatodon we recovered bat pol-
lination syndrome (PP = 0.97; Fig. 3A), open environments
(PP = 0.63 Fig. 3B), and rupicolous life form (PP = 0.67; Fig.
3C) as ancestral states.
When testing for phylogenetic correlation, the strongest eect
was the relationship between pollination syndrome and habitat
type (r = 0.39, 95% HPD -0.03 to 0.79). We also found a com-
paratively weaker relationship between pollination syndrome
and life form (Supporting Information, Material S6; r = -0.29,
95% HPD -0.66 to 0.12), representing a moderate correlation
overall. We inferred that Vriesea ancestors, as well as Stigmatodon
and Vriesea-Stigmatodon clade ones, occupied mid-elevations
(c. 700 m a.s.l.), with at least eight transitions to highlands and
ten to lowlands in Vriesea (Supporting Information, Material
S7). No correlation among pollination syndromes and altitud-
inal distribution was found (r = -0.07, 95% HPD -0.40 to 0.26,
Supporting Information, Material S7).
Distribution paerns of hummingbird and bat pollination
syndromes
Vriesea species are widely distributed throughout the latitu-
dinal and altitudinal range of the Atlantic Forest, with some of
them reaching the savanna (Brazilian Cerrado domain, in the
high campos rupestres) (Supporting Information, Material S8).
Based on our dataset, Vriesea is documented from sea level to
2162 m, between latitudes 3°S to 32°S (except for 10% of the
species occurring in the Andes, Amazon, and Greater Antilles;
Barfuss et al. 2016). High species richness is found in altitudes
up to 1200 m and between latitudes 15°S to 27°S. We found no
dierence between the occupation of latitudinal and altitudinal
space between species with hummingbird and bat pollination
Figure 2. Dated phylogenetic tree for Vriesea and Stigmatodon based on 664 ddD-seq loci estimated in treePL. Node bars indicate 95%
HPD for the age of each node. e Holocene is indicated with a bold black line. e root age of Vriesea is estimated to be placed in the Late
Miocene (6.3 Mya) and the repeated shis between pollination syndromes are estimated to occur in the Late Miocene and Early Pliocene
(5.8–5.4 Mya; blue shading indicates 95% HPD of timing of the shis). e tectonic events occurring at that time, which re-shaped the scarps
of the Serra do Mar Mountain Chain, likely increased habitat heterogeneity in which plant lineages and their mutualists diversied.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

Evolution of pollination syndromes • 7
syndromes. Instead, our results indicate that species associated
with bats reach the highest altitudes (Supporting Information,
Material S8).
DISCUSSION
e phylogenetic tree of Vriesea and its sister group Stigmatodon
using ddD-seq data revealed two independent origins of
hummingbird pollination syndrome (Fig. 1). We infer Vriesea to
have originated at 6.3 Mya and its ancestor to be a bat-pollinated
epiphyte distributed in shaded, mid-elevation areas of the
Atlantic Forest; and Stigmatodon to have originated at 8 Mya and
its ancestor to be a bat-pollinated, rupicolous species found in
open, mid-elevation granite rocky outcrops (Fig. 3; Supporting
Information, Material S7).
We have corroborated our hypotheses by showing that pol-
lination syndrome likely evolved jointly with life form and
habitat type (Supporting Information, Material S6). e in-
ferred shis from bat to hummingbird pollination correlated
with shis to epiphytism and shaded habitat, whereas bat-
pollination correlated with the rupicolous and terrestrial life
forms in open areas. Shis among syndromes were inferred
at around 5.8–5.4 Mya, during the Late Miocene and Early
Pliocene (Fig. 2). At that time, tectonic events led to geo-
logical and environmental transformations in the Serra do
Mar Mountain Chain, likely resulting in an increased habitat
heterogeneity in which plant lineages and their mutual-
ists diversied (Almeida 1976, Azevedo et al. 2020, Neves
et al. 2020b). Additionally, when investigating distribution
paerns, we showed Vriesea species from both syndromes
are distributed across wide latitudinal and altitudinal ranges
of the Atlantic Forest, reaching part of the Cerrado savanna
(Supporting Information, Material S8). To further explore
this broad occurrence paern, we leveraged Vriesea pollin-
ator interactions from compiled literature records to discuss
the inuence of pollinator variety on ecological preference
(Supporting Information, Material S8).
Pollination syndromes evolve in correlation with life form
and habitat type
We inferred a bat pollination syndrome, epiphytic life form,
and shaded habitat as the ancestral states of Vriesea, and bat-
pollinated, rupicolous life form and open habitat as ancestral
states of Stigmatodon (Fig. 3). Moreover, we showed that transi-
tions from bat to hummingbird pollination were oen linked to
transitions to the epiphytic life form in shaded habitats, whereas
bat pollination was linked to the rupicolous life form and open
habitats. Givnish et al. (2014) inferred the association between
fertile and humid mountains with epiphytism in Bromeliaceae,
suggesting that in tropical forests, abundant rainfall and the
nutrient-rich release of organic material from both plants and
animals explain the great richness of epiphytic species. Givnish
et al. (2014) also suggested such habitats favour avian pollin-
ation, especially by hummingbirds, as cool and wet conditions
select for thermoregulating pollinators (Cruden 1972, Kessler
et al. 2020). e genus Vriesea presents all these characteristics,
allowing for an understanding of the evolutionary correlates of
dierent pollination syndromes.
Figure 3. Ancestral trait estimation for (A) pollination syndromes, (B) habitat types, and (C) life forms of Vriesea and Stigmatodon. Pie charts
at nodes represent posterior probabilities of ancestral states using Bayesian inference. e most recent common ancestor of Vriesea is inferred
to have been a bat-pollinated, epiphytic plant growing in shaded forests. Shis in pollination syndrome are correlated with shis in habitat type
and life form.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

8 • Neves et al.
Despite being distributed across the entire altitudinal and
latitudinal range of the Atlantic Forest (Supporting Information,
Material S8), we identied a broad occurrence paern for
Vriesea species: hummingbird-pollinated species are oen less
exposed in shaded habitats and bat-pollinated species are oen
more exposed in open habitats. Species with owers adapted
to hummingbirds are predominant in Vriesea (c. 137 out of 230
spp.). ey are mostly epiphytes concentrated in the understory
at mid elevation, but also occur on coastal plains (restingas)
and reach high elds and rocky outcrops (campos de altitude
and campos rupestres), as terrestrial or rupicolous species and
rarely epiphytes (BFG 2018). e main hummingbird species
recorded pollinating Vriesea are the hermits of the subfamily
Phaetornithinae (P. eurynome and R. naevius, Neves et al. 2020a;
Supporting Information, Material S4). e hermits commonly
inhabit the understory of Neotropical forests and their diver-
sity decreases at high elevations and in dry habitats (Rodríguez-
Flores et al. 2019; Supporting Information, Material S9). In the
Atlantic Forest, R. naevius is amongst the main pollinators in
humid lowlands up to 500 m, while P. eurynome is predominant
at higher altitudes, around 1500 m (Buzato et al. 2000, Vizentin-
Bugoni Maruyama and Sazima 2014; Supporting Information,
Material S9). e non-hermit Vriesea pollinators of the sub-
family Trochilinae (alurania glaucopsis, Leucochloris albicolis,
Florisuga fusca Vieillot and Amazilia mbriata Gmelin) most
commonly forage in forest canopies (Supporting Information,
Material S9).
Vriesea species with owers adapted to bats are generally as-
sociated with the forest canopy, occurring as epiphytes from
low- to highlands. In the campos de altitude and campos rupestres
open elds, they usually occur as rupicolous or terrestrial spe-
cies. Due to their small size and high metabolism, nectar-feeding
bats need to quickly locate the owers to feed on using olfac-
tion, vision, and echolocation (Helversen and Winter 2003).
An experimental study conducted with the two main bat pollin-
ator species for Vriesea (A. caudifer and Anoura georoyi Gray)
showed that well-exposed owers facilitate echolocation and vi-
sion, while in a dense forest matrix, bats are more dependent on
ower scent and are guided by olfaction (Muchhala and Serrano
2015). ese two Anoura species inhabit primary and secondary
forests in Brazil, reaching altitudes up to 2000 m (Supporting
Information, Material S9). Accessibility of owers is shown
to aect bat pollination in Burmeistera H.Karst. & Triana spe-
cies, where more exposed owers have an increase in nocturnal
pollen deposition (Muchhala 2003). Dierences in ight pat-
terns of the two pollinator groups in Burmeistera could be an
explanation of the occurrence of plants in open or dense vegeta-
tion habitats, as bats sweep their wings over a large area around
their bodies while hovering, while hummingbirds are more man-
oeuvrable, with wing movements restricted to an area directly
behind their back. Here, we suggest a similar function in Vriesea,
where pollinators exert selective pressure(s) on plant habitat.
Taken together, these lines of evidence support the main occur-
rence of Vriesea hummingbird-pollinated species in shaded and
dense vegetation, and the concentration of Vriesea bat-pollinated
plants in open and exposed habitats.
We inferred the shis among syndromes to have happened
around 5.8–5.4 Mya (Fig. 2), coinciding with tectonic events of
the Late Miocene and Early Pliocene that continued shaping the
Serra do Mar Mountain Chain in the Atlantic Forest (Almeida
1976, Turcheo-Zolet et al. 2013, Guedes et al. 2020). Such
tectonic events promoted orogenic transformations likely re-
sulting in habitat barriers in which plant species and their mu-
tualists diversied. Likewise, other Atlantic Forest lineages are
hypothesized to have diverged during the same period and to be
inuenced by such changes (Grazziotin et al. 2006, Fitzpatrick
et al. 2009, omé et al. 2010). Here, we used a penalized likeli-
hood method to estimate divergence times for Vriesea instead of
a Bayesian approach as in Kessous et al. (2020), which is fast and
suitable for analysis of large datasets. However, our approach
does not account for fossil and branch length uncertainty ex-
plicitly (e.g. Reis et al. 2016). Our methodological approach in
addition to the distinct sources of information for secondary
calibrations and the inclusion of the Vriesea limae L.B.Sm.clade
as Stigmatodon (Couto et al. 2022), explain the dierences in di-
vergence times compared to Kessous et al. (2020) and Loiseau
et al. (2021). In general, our topology is congruent with other
published phylogenies for the study group, especially for deep
nodes (Kessous et al. 2020, Machado et al. 2020, Loiseau et al.
2021).
e specic habitat zones where the shis between pollin-
ation syndromes may occur are hypothesized by Kessler et al.
(2020) to be those where changes in pollinator’s physiological
preferences occur, at mid elevations and in the transitions be-
tween humid and dry areas. is hypothesis is based on a widely
recognized distribution paern of hummingbird-pollinated
species being more diverse at cool, wet, and mid to high eleva-
tions, whereas bat-pollinated species are more diverse in humid,
mid to low elevations (Kessler et al. 2020). However, this pat-
tern is recognized in studies developed along wide altitudinal
ranges, such as along the Andean slopes that reach altitudes of
more than 4000 m. In contrast, our study region—the Atlantic
Forest—only reaches altitudes of c. 2200 m, showing higher
species diversity of both hummingbird- and bat-pollinated
plant assemblages and their pollinators in the lowlands, with a
decrease of diversity towards the highlands (Sazima et al. 1999,
Buzato et al. 2000).
Species with intermediate oral morphology among the two
pollination syndromes in Vriesea are hypothesized be a product
of pollinator shis (Neves et al. 2020a). ese intermediate
species occur across the entire altitudinal range of the genus,
both in dry and wet habitats. In mutualisms that span environ-
mental gradients, specic interactions can change with biotic
and abiotic variables such as regional species guild, tempera-
ture, light, and precipitation (Chomicki et al. 2019). Broadly,
we show life form and habitat type to inuence plant–pollin-
ator interactions. Although we cannot rule out the hypoth-
esis of no correlation among such traits (considering the 95%
posterior distribution of correlation coecients, Supporting
Information, Material S6), we present corroborating evidence
from eld studies. Studies at the community level would fur-
ther clarify these factors shaping the distribution paerns of
pollination syndromes.
Repeated evolution of pollination syndromes and taxonomic
implications
We identied two independent lineages of hummingbird-
pollinated species in Vriesea (Figs 1, 3A), resulting in the repeated
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

Evolution of pollination syndromes • 9
evolution of pollination syndromes. Such ndings reinforce
the non-monophyly of Vriesea sections (V. section Vriesea and
V. section Xiphion) that were dened based on morphological
traits that reect the pollination syndromes Smith and Downs
1977.
Within each pollination syndrome, we recovered well-
supported clades of morphologically similar species (Fig. 1).
Among the hummingbird pollination syndrome are: Vriesea
teresopolitana Leme + Vriesea inata (Wawra) Wawra both from
the V. inata group, which present simple inorescences with
congested and inated oral bracts (Costa et al. 2009, Gomes-
da-Silva and Souza-Chies 2017); Vriesea guata Linden & André
and Vriesea capixabae Leme, with pendulous inorescences and
roseous bracts fully covered with a white-waxy indument (Leme
1999); Vriesea rhodostachys L.B.Sm., Vriesea a. gradata (Baker)
Mez, and Vriesea calimaniana Leme & W.Till, presenting robust
simple inorescences with large, cartaceous, very inated, and
imbricate oral bracts (Leme et al. 1997); and Vriesea sceptrum
Mez and Vriesea cacuminis L.B.Sm., both with tubular yellow
owers with included stamens, which dier from the typ-
ical hummingbird-pollinated owers in the genus. Among the
bat pollination syndrome are: Vriesea platynema Gaudich. +
Vriesea linharesiae Leme & J.A.Siqueira + Vriesea sp. with in-
cluded stamens, leaves green with a purple reddish macule in
the apex (Leme and Siqueira-Filho 2001, Moura 2011); Vriesea
roethii W.Weber and Vriesea pabstii McWill. & L.B.Sm., with
mostly green inorescences and white owers with included
stamens (Smith and Downs 1977, Weber 1979); and Vriesea
sananciscana Versieux & Wand. and Vriesea longistaminea
C.C.Paula & Leme, with exserted and slightly spread long sta-
mens, which dier from the typical bat-pollinated owers in the
genus (Moura 2011).
Despite shared morphology and, in some cases, geographic
distribution, these groupings are not exclusive for each species
complex or morphological group. Further, dierent accessions
of the same nominal species can be spread across the phylogen-
etic tree, rather than forming a monophyletic group (Fig. 1).
In addition to the poor clade resolution—resulting from the
diculty in nding informative molecular markers for Vriesea,
such ndings are suggestive of incomplete lineage sorting with
retention of ancestral polymorphism and incipient speciation
(Goetze et al. 2017). e genus is relatively young (crown age
6.3 Mya) and species may have not experienced sucient trait
or genetic coalescence. Hybridization cannot be discarded as an
alternative possible explanation, as it has been shown to occur
within Vriesea and between Vriesea and other genera (Matos et
al. 2016, Zanella et al. 2016, Neri et al. 2017, Loiseau et al. 2021).
ese studies identied breaks in reproductive isolation, such as
overlapping owering times and shared pollinators, as the main
drivers of hybridization in the group. A large phylogenetic tree of
Vriesea based on plastome data revealed similar results regarding
the non-monophyly of species (Machado et al. 2020).
Recent studies identied repeated evolution of pollination
syndromes occurring broadly in Bromeliaceae (Givnish et
al. 2014, Aguilar-Rodríguez et al. 2019) and in other diverse
tropical groups such as the Gesneriaceae and Campanulaceae
(Lagomarsino et al. 2017, Serrano-Serrano et al. 2017),
Acanthaceae (Tripp and Manos 2008), and Passioraceae
(Abrahamczyk et al. 2014). We found two shis from bat to
hummingbird pollination (Fig. 3A), whereas the reverse pat-
tern (from hummingbird to bat pollination) is more frequently
documented across angiosperms. Our ndings reject the hy-
pothesis that bat pollination is an evolutionary dead end (Tripp
and Manos 2008, Fleming et al. 2009) and corroborates evi-
dence of high transition rates from bat to hummingbird pollin-
ation (Lagomarsino et al. 2017). Although Kessler et al. (2020)
recovered hummingbird pollination as the ancestral state of
Vriesea, with three shis from hummingbird to bat pollination,
this is due to their inclusion of bat-pollinated Stigmatodon
(Barfuss et al. 2016) as Vriesea. We here follow the classication
of Barfuss et al. (2016) in accepting Stigmatodon as monophy-
letic, which is supported by the comprehensive genomic data
presented here and in Leme et al. (2017), Kessous et al. (2020),
Machado et al. (2020), and Loiseau et al. (2021).
CONCLUSION
Here we identied evolutionary correlates of plant–pollinator
interactions in an ecologically and morphologically diverse
Neotropical plant clade. Our results indicate that pollination
syndromes evolved in association with shis in plant life form
and habitat type in Vriesea and its sister group Stigmatodon.
We identied a broad paern of occurrence of hummingbird-
pollinated species in shaded and dense-forested areas, and
bat-pollinated plants in open and more exposed habitats. We
inferred bat pollination, epiphytic life form, and occupancy in
shaded and mid-elevation habitats as ancestral states in Vriesea.
e repeated evolution of pollination syndromes explains the
non-monophyly of the two Vriesea sections dened on ower
morphology. As biodiversity loss intensies globally (WWF
2020), it is crucial to understand the relationship between plants
and their pollinators to avoid erosion of the complex ecological
networks of tropical ecosystems and their capacity for mainten-
ance and continued evolution.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online
version of this article at the publisher's web-site.
Material S1. Taxa included in the phylogenetic analyses with
their respective vouchers and information on locality and origin
of the samples.
Material S2. Traits used for the ancestral character recon-
struction and correlation analyses.
Material S3. Data on pollination biology compiled by Neves
et al. (2020a).
Material S4. e full annotated ML trees of Vriesea and
Stigmatodon based on 664 ddD-seq loci inferred with both
(A) xML with BS support and (B) PhyML with aLRT SH-
like support.
Material S5. e full annotated coalescent trees of Vriesea
and Stigmatodon based on nuclear ddD-seq loci inferred with
both (A) Astral with posterior probability and (B) SVDquartets
with BS support.
Material S6. Density plot of the posterior distribution of the
correlation coecient of the Bayesian threshold models built to
test association of pollination syndromes with (A) habitat type,
(B) life form, and (C) altitude.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

10 • Neves et al.
Material S7. Ancestral trait estimation for (A) pollination
syndromes and (B) altitude.
Material S8. Vriesea species distribution across altitude and
latitude.
Material S9. Data on pollinator species distribution, habitat
preferences, and movement.
ACKNOWLEDGEMENTS
We thank Cami la Rier, Talita Machado, and Olga Kourtchenko for help
with molecular work; and Anieli Pereira, Marcos Cruz, Deise Sarzi, and
Josué Azevedo for help with analyses. e University Utrecht Botanic
Gardens and Martin-Luther-Universität Halle-Wienberg provided
samples from the live bromeliad collections. Carlos Gussoni, Roberto
Novaes, and André Siqueira provided photos and information on pollin-
ator species. is research was supported by the Museu Nacional of the
Universidade Federal do Rio de Janeiro, the International Association
for Plant Taxonomy, the Adlerbert Research Foundation (2019-410),
the Sven and Dagmar Saléns Foundation and the Wilhelm and Martina
Lundgrens Foundation (2020-3489) for the grants awarded to B.N.,
the Conselho Nacional de Desenvolvimento Cientíco e Tecnológico
- CNPq to A.F.C. (311.111/2021-1), and the Swedish Foundation
for International Cooperation in Research and Higher Education and
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—
STINT/CAPES (88881.304776.2018-01) to C.D.B. and A.F.C., the
Swedish Research Council (2017-04980) to C.D.B., the Coordenação de
Aperfeiçoamento de Pessoal de Nível Superior—CAPES to B.N., I. M.K.,
D.R.C., P.d.L.F. and R.L.M., and the Fundação de Amparo à Pesquisa
do Estado do Rio de Janeiro—FAPERJ (CNE E-26/200.940/2022) to
F.P. and I.M.K. acknowledges the Helge Ax:son Johnsons stielse (F21-
0212). A.A . acknowledges nancial support from the Swedish Research
Council (2019-05191), the Swedish Foundation for Strategic Research
(FFL15-0196), the Knut and Alice Wallenberg Foundation (W
2014.0216), and the Royal Botanic Gardens, Kew.
AUTHOR CONTRIBUTIONS
Beatriz Neves, Andrea F. Costa, Fabiano Salgueiro, Christine D.
Bacon, and Alexandre Antonelli (study design), Beatriz Neves, Igor M.
Kessous, Ricardo L. Moura, and Dayvid R. Couto (sample collection
and identication), Beatriz Neves (DNA extraction and preparation),
Francisco Prosdocimi, Paola de L. Ferreira, and Beatriz Neves (bio-
informatic analyses), Beatriz Neves, Igor M. Kessous, and Paola de L.
Ferreira (phylogenetic analyses), Beatriz Neves (remaining analyses),
all authors (data interpretation), Beatriz Neves (manuscript writing
with contributions from all authors).
DATA AVAILABILITY
Raw reads are available at NCBI under the BioProject number
PRJNA918536. e morphological and environmental data used in the
analyses are available in the electronic supplement of this paper.
CONFLICT OF INTEREST
e authors declare that they have no conict of interest.
REFERENCES
Abrahamczyk S, Souto-Vilarós D, Renner SS. Escape from ex-
treme specialization: passionowers, bats and the sword-billed
hummingbird. Proceedings of the Royal Society B: Biological Sciences
2014;281:20140888.
Aguilar-Rodríguez PA, Krömer T, Tschapka M, et al. Bat pollination in
Bromeliaceae. Plant Ecology & Diversity 2019;12:1–19.
Almeida FFM. e system of continental ris bordering the Santos Basin.
Anais da Academia Brasileira de Ciências 1976;48:15–26.
Azevedo JA, Collevai RG, Jaramillo CA, et al. On the young sa-
vannas in the land of ancient forests. In: Rull V, Carnaval AC (eds.).
Neotropical Diversication: Paerns and Processes. Cham, Springer,
2020, 271–98.
Azmat MA, Khan IA, Cheema HMN, et al. Extraction of DNA suitable
for PCR applications from mature leaves of Mangifera indica L. Journal
of Zhejiang University SCIENCE B 2012;13:239–43.
Barfuss MHJ, Till W, Leme EMC, et al. Taxonomic revision of
Bromeliaceae subfam. Tillandsioideae based on a multi-locus DNA
sequence phylogeny and morphology. Phytotaxa 2016;279:1–97.
BFG. Brazilian Flora 2020: Innovation and collaboration to meet Target
1 of the Global Strategy for Plant Conservation (GSPC). Rodriguésia
2018;69:1513–27.
Blois JL, Zarnetske PL, Fitzpatrick MC, Finnegan S. Climate change
and the past, present, and future of biotic interactions. Science
2013;341(6145):499–504.
Bollback JP. SIMMAP: stochastic character mapping of discrete traits on
phylogenies. BMC Bioinformatics 2006;7:88.
Buzato S, Sazima M, Sazima I. Hummingbird-pollinated oras at three
Atlantic Forest sites 1. Biotropica 2000;32:824–41.
Ceita GO, Assis JGA, Guedes MLS, et al. Cytogenetics of Brazilian
species of Bromeliaceae. Botanical Journal of the Linnean Society
2008;158:189–93.
Chifman J, Kubatko L. Quartet inference from SNP data under the co-
alescent. Bioinformatics 2014;3:3317–24.
Chomicki G, Weber M, Antonelli A, et al. e impact of mutualisms on
species richness. Trends in Ecology & Evolution 2019;34:698–711.
Condamine FL, Antonelli A, Lagomarsino LP, Hoorn C, Liow LH.
Teasing apart mountain upli, climate change and biotic drivers
of species diversication. In: Hoorn C, Perrigo A, Antonelli A, eds.
Mountains, Climate and Biodiversity. Hoboken: John Wiley & Sons,
2018, 257–272.
Costa AF, Gomes-da-Silva J, Wanderley MGL. Vriesea (Bromeliaceae,
Tillandsioideae): taxonomic history, and morphology of the
Brazilian lineages. e Journal of the Torrey Botanical Society
2014;141:338–52.
Costa AF, Gomes-da-Silva J, Wanderley MGL. Vriesea (Bromeliaceae,
Tillandsioideae): a cladistic analyses of eastern Brazilian species based
on morphological characters. Rodriguésia 2015;66:429–40.
Costa AF, Rodrigues PJFP, Wanderley MGL. Morphometric analysis of
Vriesea paraibica Wawra complex (Bromeliaceae). Botanical Journal of
the Linnean Society 2009;159:163–81.
Cotias-de-Oliveira ALP, Assis JGA, Ceita G, et al. Chromosome
number for Bromeliaceae species occurring in Brazil. Cytologia
2004;69:161–6.
Couto DR, Kessous IM, Neves B, et al. Molecular phylogenetics and
character evolution in Stigmatodon (Bromeliaceae, Tillandsioideae),
an endemic genus to Brazilian rocky outcrops. Systematic Botany
2022;47:347–362.
Crepet WL, Niklas KJ. Darwin’s second ‘abominable mystery’: Why
are there so many angiosperm species? American Journal of Botany
2009;96:366–81.
Cruden RW. Pollinators in high-elevation ecosystems: relative eective-
ness of birds and bees. Science 1972;176:1439–40.
Dellinger AS. Pollination syndromes in the 21st century: where do we
stand and where may we go? New Phytologist 2020;228:1193–213.
Doyle JJ, Doyle JL. A rapid DNA isolation procedure for small quantities
of fresh leaf tissue. Phytochemical Bulletin 1987;19:11–5.
Eaton DAR. PyD: assembly of de novo Dseq loci for phylogenetic
analyses. Bioinformatics 2014;30:1844–9.
Eaton DAR, Ree RH. Inferring phylogeny and introgression using
Dseq data: an example from owering plants (Pedicularis:
Orobanchaceae). Systematic Biology 2013;62:689–706.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

Evolution of pollination syndromes • 11
Faegri K, van der Pijl L. e Principles of Pollination Ecology. Oxford:
Pergamon Press, 1979.
Felsenstein J. A comparative method for both discrete and continuous
characters using the threshold model. e American Naturalist
2012;179:145–56.
Fenster CB, Armbruster WS, Wilson P, et al. Pollination syndromes
and oral specialization. Annual Review of Ecology, Evolution, and
Systematics 2004;35:375–403.
Ferreira PL, Batista R, Andermann T, et al. Target sequence capture
of Barnadesioideae (Compositae) demonstrates the utility of low
coverage loci in phylogenomic analyses. Molecular Phylogenetics and
Evolution 2022;169:107432.
Fitzpatrick SW, Brasileiro CA, Haddad CFB, et al. Geographical vari-
ation in genetic structure of an Atlantic Coastal Forest frog re-
veals regional dierences in habitat stability. Molecular Ecology
2009;18:2877–96.
Fleming TH, Geiselman C, Kress WJ. e evolution of bat pollination: a
phylogenetic perspective. Annals of Botany 2009;104:1017–43.
Givnish TJ, Barfuss MHJ, Ee BV, et al. Adaptive radiation, correlated and
contingent evolution, and net species diversication in Bromeliaceae.
Molecular Phylogenetics and Evolution 2014;71:55–78.
Goetze M, Zanella CM, Palma-Silva C, et al. Incomplete lineage sorting
and hybridization in the evolutionary history of closely related, en-
demic yellow-owered Aechmea species of subgenus Ortgiesia
(Bromeliaceae). American Journal of Botany 2017;104:1073–1087.
Gomes-da-Silva J, Souza-Chies . What actually is Vriesea? A total evi-
dence approach in a genus of Tillandsioideae (Bromeliaceae, Poales).
Cladistics 2017;34:181–99.
Gouda EJ, Butcher D, Gouda K . (continuously updated). Encyclopaedia of
Bromeliads. Available at: hps://bromeliad.nl/encyclopedia/
Grant JR. Bromelienstudien. e resurrection of Alcantarea and Werauhia,
a new genus. Tropische und Subtropische Panzenwelt 1995;91:1–57.
Grazziotin FG, Monzel M, Echeverrigaray S, et al. Phylogeography of the
Bothrops jararaca complex (Serpentes: Viperidae): Past fragmenta-
tion and island colonization in the Brazilian Atlantic Forest. Molecular
Ecology 2006;15:3969–82.
Guedes TB, Azevedo JAR, Bacon CD, Provete DB, Antonelli A. Diversity,
endemism, and evolutionary history of montane biotas outside the
Andean region. In: Rull V, Carnaval A, eds. Neotropical Diversication:
Paerns and Processes. Chamonix: Springer, 2020, 299–328.
Guindon S, Dufayard JF, Lefort V, et al. New algorithms and methods to
estimate maximum-likelihood phylogenies: assessing the perform-
ance of PhyML 3.0. Systematic Biology 2010;59:307–21.
Helversen OV, Winter Y. Glossophagine bats and their owers: costs and
benets for plants and pollinators. In: Kunz TH, Fenton MB, eds. Bat
Ecology. Chicago: University of Chicago Press, 2003, 346–397.
Hillis DM, Bull JJ. An empirical test of bootstrapping as a method for
assessing condence in phylogenetic analysis. Systematic Biology
1993;42:182–92.
Kessler M, Abrahamczyk S, Krömer T. e role of hummingbirds in the
evolution and diversication of Bromeliaceae: unsupported claims
and untested hypotheses. Botanical Journal of the Linnean Society
2020;192:592–608.
Kessous IM, Neves B, Couto DR, et al. Historical biogeography
of a Brazilian lineage of Tillandsioideae (subtribe Vrieseinae,
Bromeliaceae): e Paranaean Sea hypothesized as the main vicariant
event. Botanical Journal of the Linnean Society 2020;192:625–41.
Lagomarsino L, Condamine F, Antonelli A, et al. e abiotic and biotic
drivers of rapid diversication in Andean bellowers. New Phytologist
2016;210:1430–42.
Lagomarsino LP, Forrestel EJ, Muchhala N, et al. Repeated evolution of
vertebrate pollination syndromes in a recently diverged Andean plant
clade. Evolution 2017;71:1970–85.
Langmead B, Salzberg S. Fast gapped-read alignment with Bowtie 2.
Nature Methods 2012;9:357–9.
Leaché AD, Fujita MK, Minin VN, et al. Species delimitation using
genome-wide SNP data. Systematic Biology 2014;63:534–42.
Leme EM, Halbrier H, Barfuss MH. Waltillia, a new monotypic genus in
Tillandsioideae (Bromeliaceae) arises from a rediscovered, allegedly
extinct species from Brazil. Phytotaxa 2017;299:1–35.
Leme EM, Siqueira-Filho JA. Studies in Bromeliaceae of northeastern
Brazil. I. Selbyana 2001;22:146–54.
Leme EMC. New species of Brazilian Bromeliaceae: a tribute to Lyman B.
Smith. Harvard Papers in Botany 1999;4:135–67.
Leme EMC, Halbrier H, Gortan G. A new species of Vriesea from
Espírito Santo. Bromélia 1997;4:7–12.
Lexer C, Marthaler F, Humbert S, et al. Gene ow and diversication in a
species complex of Alcantarea inselberg bromeliads. Botanical Journal
of the Linnean Society 2016;181:505–20.
Loiseau O, Machado TM, Paris M, et al. Genome skimming reveals
widespread hybridization in a Neotropical owering plant radiation.
Frontiers in Ecology and Evolution 2021;9:668281.
Machado TM, Loiseau O, Paris M, et al. Systematics of Vriesea
(Bromeliaceae): phylogenetic relationships based on nuclear gene
and partial plastome sequences. Botanical Journal of the Linnean
Society 2020;192:656–74.
Manhães VC. Diversidade Morfológica e Genética em Stigmatodon
(Tillandsioideae, Bromeliaceae). Unpublished D. Phil. esis, Museu
Nacional, Universidade Federal do Rio de Janeiro, 2021.
Massai R, Reznicek AA, Knowles LL. Utilizing Dseq data for
phylogenetic analysis of challenging taxonomic groups: a case
study in Carex sect. Racemosae. American Journal of Botany
2016;103:337–47.
Matos JZ, Juan A, Agullo JC, et al. Morphological features, nuclear
microsatellites and plastid haplotypes reveal hybridization processes
between two sympatric Vriesea species in Brazil (Bromeliaceae).
Phytotaxa 2016;261:58–74.
Miller MA, Pfeier W, Schwartz T. Creating the CIPRES science gateway
for inference of large phylogenetic trees. In: 2010 Gateway Computing
Environments Workshop. New Orleans: IEEE, 1–8, 2010
Moura RL. Revisão taxonômica do grupo Vriesea platynema Gaudich.
(Bromeliaceae). Unpublished D. Phil. esis, Museu Nacional,
Universidade Federal do Rio de Janeiro, 2011.
Muchhala N. Exploring the boundary between pollination syndromes:
bats and hummingbirds as pollinators of Burmeistera cyclostigmata and
B. tenuiora (Campanulaceae). Oecologia 2003;134:373–80.
Muchhala N, Serrano D. e complexity of background cluer af-
fects nectar bat use of ower odor and shape cues. PLoS ONE
2015;10:e0136657.
Murrell P. 2018. R graphics. London: Chapman & Hall/CRC.
Myers N, Miermeier , Miermeier CG, et al. Biodiversity hotspots
for conservation priorities. Nature 2000;403:853–8.
Neri J, Wendt T, Palma-Silva C. Natural hybridization and genetic and
morphological variation between two epiphytic bromeliads. AoB
Plants 2017;10:plx061.
Neves B, Kessous IM, Moura RL, et al. Pollinators drive oral evolution in
an Atlantic Forest genus. AoB Plants 2020a;12:plaa046.
Neves B, Uribbe FP, Jacques SSA, et al. Species boundaries in the Vriesea
incurvata (Bromeliaceae) complex aer a broad morphometric and
taxonomic study. Systematic Botany 2018;43:870–88.
Neves B, Zanella CM, Kessous IM, et al. Drivers of bromeliad leaf and
oral bract variation across a latitudinal gradient in the Atlantic
Forest. Journal of Biogeography 2020b;47:261–74.
Nogueira MGC. Vriesea Lindl. (Bromeliaceae, Tillandsioideae) Para a
Flora da Bahia: Descrições, Novas Ocorrências e Dois Novos Sinônimos.
Unpublished Master’s esis, Universidade Estadual de Feira de
Santana, 2013.
Ollerton J, Alarcón R, Waser NM, et al. A global test of the pollination
syndrome hypothesis. Annals of Botany 2009;103:1471–80.
Palma-Silva C, dos Santos DG, Kaltchuk-Santos E, et al. Chromosome
numbers, meiotic behavior, and pollen viability of species of Vriesea
and Aechmea genera (Bromeliaceae) native to Rio Grande do Sul,
Brazil. American Journal of Botany 2004;91:804–7.
Peterson BK, Weber JN, Kay EH, et al. Double digest Dseq: an inex-
pensive method for de novo SNP discovery and genotyping in model
and non-model species. PLoS ONE 2012;7:e37135.
Plummer M, Best N, Cowles K, et al. CODA: convergence diagnosis and
output analysis for MCMC. R News 2006;6:7–11.
Rambaut A. FigTree 1.4.2 soware. Edinburgh: Institute of Evolutionary
Biology, University of Edinburgh, 2014.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023

12 • Neves et al.
Ramos FN, Mortara SR, Monalisa-Francisco N, et al. Atlantic epiphytes: a
data set of vascular and non-vascular epiphyte plants and lichens from
the Atlantic Forest. Ecology 2019;100:e02541.
R Core Team. R: A Language and Environment for Statistical Computing.
Vienna: R Foundation for Statistical Computing. 2018.
Reis M, Donoghue PC, Yang Z. Bayesian molecular clock dating of
species divergences in the genomics era. Nature Reviews Genetics
2016;17:71–80.
Revell LJ. phytools: an R package for phylogenetic comparative
biology (and other things). Methods in Ecology & Evolution
2012;3:217–23.
Rodríguez-Flores CI, Ornelas JF, Wethington S, et al. Are hummingbirds
generalists or specialists? Using network analysis to explore the mech-
anisms inuencing their interaction with nectar resources. PLoS ONE
2019;14:e0211855.
Rosas-Guerrero V, Aguilar R, Martén-Rodríguez S, et al. A quantitative
review of pollination syndromes: do oral traits predict eective pol-
linators? Ecology Leers 2014;17:388–400.
Sazima M, Buzato S, Sazima I. Bat pollination of Vriesea in southeastern
Brazil. Bromelia 1995;2:29–37.
Sazima M, Buzato S, Sazima I. Bat-pollinated ower assemblages and
bat visitors at two Atlantic Forest sites in Brazil. Annals of Botany
1999;83:705–12.
Serrano-Serrano ML, Rolland J, Clark JL, et al. Hummingbird pollination
and the diversication of angiosperms: an old and successful asso-
ciation in Gesneriaceae. Proceedings of the Royal Society B: Biological
Sciences 2017;284:20162816e20162816.
Silvestro D, Zizka G, Schulte K. Disentangling the eects of key in-
novations on the diversication of Bromelioideae (Bromeliaceae).
Evolution 2014;68:163–75.
Smith LB, Downs RJ. Tillandsioideae (Bromeliaceae). Flora Neotropica
Monograph. New York: Hafner Press, 1977.
Smith SA, O’Meara BC. treePL: divergence time estimation using
penalized likelihood for large phylogenies. Bioinformatics
2012;28:2689–90.
Stamatakis A. RAxML version 8: a tool for phylogenetic ana-
lysis and post-analysis of large phylogenies. Bioinformatics
2014;30:1312–3.
Suchard MA, Lemey P, Baele G, et al. Bayesian phylogenetic and
phylodynamic data integration using BEAST 1.10. Virus Evolution
2018;4:vey016.
omé MTC, Zamudio KR, Giovanelli JG, et al. Phylogeography of en-
demic toads and post-Pliocene persistence of the Brazilian Atlantic
Forest. Molecular Phylogenetics and Evolution 2010;55:1018–31.
Tripp EA, Manos PS. Is oral specialization an evolutionary dead-end?
Pollination system transitions in Ruellia (Acanthaceae). Evolution
2008;62:1712–37.
Turcheo-Zolet AC, Pinheiro F, Salgueiro F, et al. Phylogeographical pat-
terns shed light on evolutionary process in South America. Molecular
Ecology 2013;22:1193–213.
Uribbe FP, Neves B, Jacques SSA, et al. Morphological variation in Vriesea
procera complex (Bromeliaceae, Tillandsioideae) in Brazilian Atlantic
Rainforest, with recognition of new taxa. Systematic B otany 2020;45:53–68.
Van der Niet T, Johnson SD. Phylogenetic evidence for pollinator-
driven diversication of angiosperms. Trends in Ecology & Evolution
2012;27:353–61.
Versieux LM, Wanderley MGL. A new species of Vriesea Lindl.
(Bromeliaceae, Tillandsioideae) from Serra da Canastra, Minas
Gerais State, Brazil. Acta Botanica Brasilica 2008;22:71–4.
Vizentin-Bugoni J, Maruyama PK, Sazima M. Processes entangling inter-
actions in communities: forbidden links are more important than
abundance in a hummingbird–plant network. Proceedings of the Royal
Society B: Biological Sciences 2014;281:20132397.
Weber W. Species Novae Bromeliacearum II. Bradea 1979;3:23.
WFO (World Flora Online). hp://www.worldoraonline.org. (13
September 2020, date last accessed).
Wickham H. ggplot2. Wiley Interdisciplinary Reviews Computational
Statistics 2011;3:180–5.
WWF. Living Planet Report 2020 - Bending the Curve of Biodiversity Loss,
Gland, Switzerland: WWF 2020.
Zanella CM, Palma-Silva C, Goetze M, et al. Hybridization between
two sister species of Bromeliaceae: Vriesea carinata and V. i n c u rv a t a .
Botanical Journal of Linnean Society 2016;181:491–504.
Zhang C, Rabiee M, Sayyari E, et al. ASTL-III: polynomial time
species tree reconstruction from partially resolved gene trees. BMC
Bioinformatics 2018;19:153.
Downloaded from https://academic.oup.com/botlinnean/advance-article/doi/10.1093/botlinnean/boad015/7216834 by guest on 04 July 2023