vol. 194, no.

1 the american naturalist july 2019

Natural History Note

Bimodal Pollination Systems in Andean Melastomataceae

Involving Birds, Bats, and Rodents

Agnes S. Dellinger,1,* Lisa M. Scheer,2 Silvia Artuso,2 Diana Fernández-Fernández,3

Francisco Sornoza,3 Darin S. Penneys,4 Raimund Tenhaken,2 Stefan Dötterl,2
and Jürg Schönenberger1

1. Department of Botany and Biodiversity Research, University of Vienna, Rennweg 14, 1030 Vienna, Austria; 2. Department of Biosciences,
University of Salzburg, Hellbrunnerstrasse 34, 5020 Salzburg, Austria; 3. Herbario Nacional del Ecuador, Instituto Nacional de Biodiversidad,
Avenida Río Coca E6-115 e Isla Fernandina, Quito, Ecuador; 4. Department of Biology and Marine Biology, University of North Carolina
Wilmington, Wilmington, North Carolina 28403
Submitted June 27, 2018; Accepted February 13, 2019; Electronically published May 23, 2019
Online enhancements: appendix, videos. Dryad data: https://dx.doi.org/10.5061/dryad.jk673fq.

abstract: Floral adaptation to a single most effective functional pol-

tion for adaptation to a plant species’ most effective pollina-
linator group leads to specialized pollination syndromes. However, tor (Stebbins 1970; Fenster et al. 2004). Pollinator effective-
adaptations allowing for pollination by two functional groups (bi- ness is generally understood as the product of pollinator
modal pollination systems) remain a rarely investigated conundrum. “quantity” (visitation frequency) and “quality” (efficiency
We tested whether floral scent and nectar traits of species visited by in conspecific pollen transfer). These ideas are essential in
two functional pollinator groups indicate specialization on either of the concept of pollination syndromes, assuming convergent
the two pollinator groups or adaptations of both (bimodal systems). We
floral evolution in adaptation to a specific (most effective)
studied pollination biology in four species of Meriania (Melastomata-
ceae) in the Ecuadorian Andes. Pollinator observations and exclusion ex-
functional pollinator group (Faegri and van der Pijl 1979;
periments showed that each species was effectively pollinated by two Fenster et al. 2004; Rosas-Guerrero et al. 2014). Although
functionalgroups(hummingbirds/bats, hummingbirds/rodents,flower- specialization on the most effective pollinator is generally
piercers/rodents), nectar composition followed known bird preferences, assumed, generalization by floral adaptation to several rel-
and scent profiles gave mixed support for specialization on bats and atively ineffective pollinators in addition to the most effec-
rodents. Our results suggest that nectar-rewarding Meriania species have tive pollinator can evolve if this results in an overall fitness
evolved stable bimodal pollination strategies with parallel adaptations to
gain (Aigner 2001, 2006) and if pollinator-mediated adap-
two functional pollinator groups. The discovery of rodent pollination is
particularly important given its rarity outside of South Africa.
tive trade-offs are minimal (Thomson 2003; Castellanos
et al. 2004; Muchhala 2007). Bimodal pollination systems,
Keywords: mixed pollination systems, buzz pollination, nectar, floral defined as systems effectively pollinated by two different
scent, pollinator effectiveness, rodent pollination. functional groups and intermediate in adaptation between
two pollination syndromes, are particularly interesting in
the specialist-generalist continuum (Manning and Goldblatt
Introduction
2005).
Specialization between plants and pollinators is regarded as In this context, it is important to distinguish between bi-
an integral process in angiosperm evolution driven by selec- modal pollination systems with two almost equally effec-
tive pollinators and pollination systems with one primary
(most effective) and one secondary (less effective) pollina-
* Corresponding author; email: [email protected]. tor (Waser 1996; Manning and Goldblatt 2005). The latter
ORCIDs: Dellinger, https://orcid.org/0000-0003-1394-3414; Tenhaken, https:// often includes a plant lineage’s ancestral, presently quite
orcid.org/0000-0002-6764-4826; Schönenberger, https://orcid.org/0000-0001-6791
ineffective pollinator (Rosas-Guerrero et al. 2014). In the
-2731.
New World tropics, for example, a small number of mixed
Am. Nat. 2019. Vol. 194, pp. 104–116. q 2019 by The University of Chicago. hummingbird-bat systems have been described (e.g., Much-
0003-0147/2019/19401-58572$15.00. All rights reserved. This work is licensed
hala et al. 2009 and references therein; Amorim et al. 2013;
under a Creative Commons Attribution-NonCommercial 4.0 International
License (CC BY-NC 4.0), which permits non-commercial reuse of the work with
Queiroz et al. 2016). Bat pollination is believed to have
attribution. For commercial use, contact [email protected]. evolved from hummingbird pollination in these systems
DOI: 10.1086/703517 (Rosas-Guerrero et al. 2014; but see also Fleming et al.
 Mixed Pollination Systems in Meriania 105

2009). Hence, either hummingbirds could represent ances- combinations of pollinators; instead, results suggested pool-
tral, less effective pollinators or these systems could consti- ing them into a single “mixed vertebrate” syndrome (Del-
tute bimodal pollination systems, with adaptations to both linger et al. 2019a). This syndrome is characterized by the
bats and hummingbirds. visitors’ shared interest in the nectar reward and their abil-
Given obvious differences in morphology, activity pat- ity to cause pollen release via a “saltshaker” mechanism.
terns, and sensory abilities between birds and bats, these Pollen release is activated when the visitors insert their
systems are ideal for testing hypotheses on floral adaptation mouthparts into the pendant, pseudocampanulate corol-
and pollinator effectiveness. Crucial features of (bimodal) las to take up nectar and thereby push against the thecae
hummingbird-bat systems would include both diurnal and (fig. 1A–1D, 1I, 1J). It remains unclear, however, whether
nocturnal anthesis, attractor cues for and morphological fit the different functional pollinator groups are equally effec-
with both pollinator groups, and accessibility and continu- tive and whether the mixed vertebrate syndrome actually
ous availability of nectar rewards (Muchhala et al. 2009). comprises a variety of truly bimodal systems. Alternatively,
For example, visual attractiveness is generally associated these systems could represent transitions (shifts) between
with diurnal pollination (e.g., red corollas in bird systems; two functional pollinator groups that differ in pollination
Faegri and van der Pijl 1979). Floral scent, on the other effectiveness.
hand, is regarded as a particularly important attractant for In this study, we selected four Meriania species of the
nocturnally active pollinators and is less important in diur- mixed vertebrate syndrome to test differences in pollinator
nal bird pollination systems (Raguso et al. 2003; Dobson efficiency by assessing quantity (visitation rate) and qual-
2006; but note that scent can also function as a deterrent, ity (in terms of pollen deposition on stigmas) of diurnal
e.g., of herbivores). Furthermore, specific scent bouquets and nocturnal pollinators. We demonstrate that nectar re-
have been related to different functional pollinator groups wards are easily accessible to all functional pollinator groups
in some plant lineages (Knudsen and Tollsten 1995; Dobson involved and test whether nectar and scent composition
2006; Knudsen et al. 2006). Strong associations between show adaptations to a single pollinator group or adaptations
nectar sugars and pollinator group/specificity have also been for bimodal pollination systems.
documented (e.g., Baker and Baker 1983; Dupont et al. 2004;
Johnson and Nicolson 2008). Flowers pollinated by special-
ized nectar-feeding birds, for example, present nectar rich
in sucrose, while hexose dominates in nectar of generalist Methods
bird-pollinated flowers (Baker et al. 1998; Johnson and Nic-
Taxon Sampling and Study Design
olson 2008). Bat-pollinated flowers in the New World are in-
termediate in sucrose and hexose levels (Abrahamczyk et al. The four selected Meriania species stem from two inde-
2017). pendent shifts from ancestral bee-buzz pollination to alter-
The plant family Melastomataceae (ca. 5,000 sp.) is func- native pollinators (shift 1: Meriania furvanthera, Meriania
tionally specialized on bee-buzz pollination and character- tomentosa; shift 2: Meriania aff. sanguinea, Meriania sangui-
ized by nectarless flowers, anthers opening by small apical nea; Dellinger et al. 2019a). The exact taxonomic status of
pores, and pollen as sole reward (Buchmann 1983; Renner M. aff. sanguinea is unclear; this taxon occurs in an isolated
1989). However, nectar secretion and concomitant pollina- population in northern Ecuador, while M. sanguinea is re-
tor shifts from pollen-collecting bees to nonbuzzing insects stricted to southern Ecuador and northern Peru (Wurdack
or vertebrates have been documented in ca. 100 Neotropical 1967). The northern population has generally been treated
Melastomataceae species scattered across four tribes (e.g., as M. sanguinea, but given clear morphological and molec-
Lumer 1980; Vogel 1997; Kriebel and Zumbado 2014; Wester ular differences (Dellinger et al. 2019a), we treat it as a sep-
et al. 2016; Brito et al. 2017). Although ambiguity remains as arate taxon in this study.
to where and how nectar is secreted (Stein and Tobe 1989; Meriania species are shrubs or treelets, mostly growing in
Varassin et al. 2008), the shift from pollen to nectar reward- small, isolated populations in montane rainforests (1,500–
ing clearly opened up the specialized bee-buzz pollination 3,200 m) of the tropical Andes, the world’s richest biodi-
syndrome to multiple functional pollinator groups (Brito versity hotspot (Myers et al. 2000). Extensive field studies
et al. 2017; Dellinger et al. 2019a). In the tribe Merianieae, were conducted in Ecuador in October/November 2016
for example, all investigated nectar-rewarding species were and 2017 (M. aff. sanguinea: Guanderas Reserve; M. furvan-
visited by different combinations of one diurnal and one thera and M. sanguinea: Podocarpus National Park; M.
nocturnal functional pollinator group (e.g., hummingbirds tomentosa: Bellavista Reserve). We aimed at locating the
and bats or hummingbirds and rodents; Dellinger et al. maximum number of accessible flowering individuals along
2019a). Using multivariate analyses of floral traits, it was different trails at each forest site, with the total sampling
impossible to distinguish flowers based on these different area spanning a minimum air line distance of 500 m at
106 The American Naturalist

Figure 1: Inflorescences, flowers, and nectar secretion in Meriania species of the mixed vertebrate syndrome. A, Meriania tomentosa with
protruding inflorescence and flame-throated sunangel visiting a flower. B, Meriania furvanthera with flowers arranged in a simple dichasium,
allowing flowerpiercers and rodents to perch close to flowers. C, Multiflowered inflorescence on a procumbent branch of Meriania sanguinea,
allowing access for hummingbirds and rodents; arrowheads indicate site of nectar aggregation. D, Fully anthetic flower of M. tomentosa with
reflexed stamens, pores, and stigma positioned at corolla opening; arrowheads indicate location of nectar aggregation. E, M. tomentosa,
anthetic flower seen from the side with petals partly removed, showing dorsal side of filaments with ruptures secreting nectar (arrowhead).
F, Nectar drop (arrowhead) on filament ruptures in M. tomentosa (type a). G, Stamens of M. furvanthera with nectar visible on ventral side of
filament-connective joint (arrowheads; types b, c). H, Generalized schematic drawing of a nectar-secreting Meriania flower at the beginning
of anthesis; stamen is bent with anther tip pointing toward the style base; no nectar secretion yet. I, Schematic drawing of an anthetic M.
tomentosa flower; stamens are erect with the anther tip and the pore close to the stigma; nectar-secreting filament ruptures are indicated
(type a); shaded area indicates position of nectar aggregation on corolla. J, Schematic drawing of an anthetic M. sanguinea flower; stamens
are erect and the anthers are distinctly curved; anther tip is close to the stigma; arrowhead indicates location of nectar secretion on ventral
side of filament-connective joint (type b); shaded area indicates position of nectar aggregation on corolla. Type of nectar secretion is given as
described in the main text. c p corolla; f p filament; h p hypanthium; p p pore; s p style.

each site to buffer known effects of small-scale differences Pollinator Quantity and Quality
in pollinator activity (e.g., Akter et al. 2017; number of in-
dividuals studied: M. aff. sanguinea: 7; M. furvanthera: 3; To assess visitation rates (quantity), flowers of multiple in-
M. sanguinea: 19; M. tomentosa: 7; for details, see table A1; dividuals (two to 10) per species were monitored using
tables A1–A13 are available online). video cameras (Sony HDR-CX190 camcorder; table A3 de-
 Mixed Pollination Systems in Meriania 107

tails sample sizes). Cameras were placed on tripods approx- tive other time interval (daytime visitors excluded from
imately 2 m away from the plants, and single inflorescences 05:45 to 18:00; nighttime visitors excluded from 18:00 to
were filmed during daytime (06:00–18:00) and nighttime 05:45; for details on sample sizes, see table A4; total flowers,
(18:00–00:00). In each video, a minimum of three 30-min n p 80). From day 1 to day 4, consecutively opening virgin
intervals (beginning/middle/end of video) were replayed, flowers within each inflorescence were added to the expo-
yielding a total of 108 reviewed hours (table A3). Floral sure trials; flowers opening on days 5–7 were not consid-
visitors were scored as pollinators if their morphology fitted ered. After 3 days or nights of pollinator exclusion (which
with the flower and their behavior could cause pollen ejec- also marks the end of the flower’s life span), styles were col-
tion. Visitation rates were calculated as pollinator visit per lected in 70% ethanol. Unmanipulated control flowers from
flower per hour (table 1). Most inflorescences presented inflorescences not used in exclusion experiments or neigh-
more than one open flower so that it was possible to mon- boring individuals (when not enough inflorescences were
itor multiple flowers simultaneously (yielding a total of more present on individuals used for exclusion trials; table A4)
than 390 flower observation hours; for a similar approach, were used to assess stigma pollen loads under natural con-
see Muchhala et al. 2009). Pollinators were identified with ditions. In the lab, stigmas were cut from styles, placed into
the help of literature (Ridgely and Greenfield 2001; Tirira a drop of lactic acid on microscope slides, squashed with a
2017). coverslip to flatten out the tissue, and viewed under a fluo-
To understand whether the different functional pollina- rescence microscope (Kearns and Inouye 1993). The entire
tor groups differed in their quality (measured by pollen de- squashed stigmatic area was measured at #10 magnifica-
position on stigmas), we conducted a pollinator exclusion tion, and pollen grains were counted at #20 (entire field
experiment over a 7-day period in M. aff. sanguinea, M. of view) in three areas from the edge to the center of the
sanguinea, and M. tomentosa (tables A1, A4; M. furvan- stigma. Pollen grain sizes of all species had been measured
thera was excluded from this experiment as too few individ- previously (17.3–19.9 mm), and pollen grains of sizes differ-
uals with accessible flowers were available). As each species ent from those of Meriania were excluded from counting.
was visited by one functional group during daytime and Total pollen grain number was calculated by multiplying
another functional group during nighttime, we excluded total stigma area by mean pollen grain number per mm2.
either daytime or nighttime visitors by bagging inflores- For each species, a generalized linear mixed effects model
cences with bridal veil (mesh density ! 1 mm) at the respec- (GLMM) was used to test for differences between diurnal

Table 1: Pollinator assemblages and visitation rates per flower per hour of the four Meriania species and total number
of flower observation hours in parentheses
Diurnal Nocturnal
visitation Nocturnal visitation
Species Group Diurnal pollinators rate/flower/hour pollinators rate/flower/hour
Meriania aff. sanguinea HB Eriocnemis derbyi (black- 1.7 (50) Anoura cf. .88 (45.2)
thighed puffleg); peruviana (bat)
Metallura tyrianthina
(Tyrian metaltail); Schistes
geoffroyi (wedge-billed
hummingbird)
Meriania tomentosa HB Coeligena torquata (collared 3.49 (15.5) Anoura sp. (bat) .73 (20.5)
(Cogn.) Wurdack inca); Adelomyia
melanogenys (speckled
hummingbird); Ocreatus
underwoodii (booted
racket-tail); Urosticte
benjamini (purple-bibbed
whitetip)
Meriania sanguinea HR Heliangelus micraster .25 (52) Thomasomys sp. .03 (63.7)
Wurdack (flame-throated sunangel) (rodent)
Meriania furvanthera FR Diglossa cyanea (masked .11 (94) Rodent, unidentified .46 (43.6)
Wurdack flowerpiercer)
Note: For details, see table A3, available online. Pollinator groups: FR p flowerpiercer/rodent; HB p hummingbird/bat; HR p hummingbird/rodent.
108 The American Naturalist

and nocturnal stigma pollen loads and between controls furvanthera, the nectar volume could not be measured due
and exclusion trials, including plant individual as a random to small sample sizes, and the concentration was measured
effect (lmerTest package in R; Kuznetsova et al. 2017). We from unbagged flowers. Summary statistics were calculated
can rule out possible confounding effects of pollen deposi- for all species from all measurements. GLMMs were used
tion on stigmas by our own disturbance when bagging/un- to assess significant differences in nectar concentration and
bagging inflorescences, as pollen is retained within the pori- volume between day and night measurements in M. tomen-
cidal anthers. Also, unmanipulated control flowers did not tosa and M. sanguinea, setting treatment (day/night) as a
show significantly different stigmatic pollen loads than the fixed factor and flower ID as a random effect (M. aff. san-
median diurnal plus nocturnal stigmatic pollen loads from the guinea was excluded due to small n). This model was chosen
exclusion experiment (Kruskal-Wallis x2: 0.428, df p 1, to specifically account for the effects of repeated sampling
P p :513). of individual flowers. Another GLMM was run on all mea-
surements on nectar concentration (n p 105) to assess sig-
nificant differences between species and day/night, treating
Localization of Nectaries plant individual as a random effect to account for the re-
peated sampling of different flowers on the same plant indi-
To provide a better understanding of the evolution of nectar-
vidual.
rewarding flowers from pollen-rewarding ancestors in the
Ten microliters of the nectar collected at day/night sam-
family, we compared nectar-secreting structures of the four
pling times was stored in 70% ethanol for sugar analyses
study species plus six additional nectar-secreting species (ta-
using high-performance liquid chromatography (HPLC; a
ble A2). Note that there is no underlying expectation related
total of 87 samples; table A6). Nectar sugar samples were
to nectar-secreting strategies and the different mixed pol-
dried in a vacuum concentrator centrifuge to remove etha-
linator assemblages. Ethanol-preserved floral material was
nol and redissolved in 500 mL of water. For HPLC, an ali-
studied with SEM or LM to localize areas of nectar secre-
quot from each sample was further diluted 1∶100 with
tion. For SEM, hypanthia and stamens were dehydrated over
water and analyzed on an ICS300 HPLC (Dionex) using
an ethanol series, transferred to acetone, critical-point dried
anion-exchange chromatography coupled with pulsed am-
(Autosamdri-815), mounted on stubs, sputter-coated with
perometric detection. Sugars were separated on a CarboPac
gold (SCD 050), and scanned in a JEOL JSM-6390 at 10 kV.
PA1 column (two 250-mm separation columns and two
For producing serial thin sections, material was dehydrated,
50-mm guard columns) using isocratic separation with
infiltrated (Technovit 7100, hardener I; Heraeus Kulzer,
80 mM NaOH and a flow rate of 0.25 mL min21. Authentic
Wehrheim, Germany), embedded in 2-hydroxyethyl meth-
standards were separated for calibration to ensure proper
acrylate (Technovit 7100, hardener II), and sectioned at 5 mm
quantification of each sugar. For each sample, the percent-
with a Microm HM rotary microtome 355 (Walldorf, Ger-
age of glucose, fructose, and sucrose was calculated for day
many). Sections were stained with 0.2% ruthenium red and
and night (Baker and Baker 1983). Bray-Curtis dissimilarity
0.5% toluidine (RT stain). Images of selected sections were
matrices were calculated in R package vegan (Oksanen et al.
taken with a Nikon digital sight DS-Fi1 camera (Nikon, Tokyo)
2018), and a two-way crossed permutational MANOVA
on an Olympus BX50 system microscope (Olympus, Tokyo).
(PERMANOVA) was run with pairwise comparison and
a Bonferroni correction to test for significant differences
in nectar composition between species, day/night, and indi-
Nectar Collection and Analyses
viduals (pairwiseAdonis; Martinez Arbizu 2017). Disparity
To assess differences in nectar properties between day and in sugar composition was calculated (betadisper function),
night, flowers of any age were bagged in the early morning and ANOVA was used to test for significant differences in
(05:30–07:00) or early evening (17:30–18:45) after remov- disparity between species.
ing all nectar if present (table A6). The age of a flower was
scored as “first day,” “second day,” or “old” by the degree
Volatile Collection and Analyses
of petal spreading and anther reflexion to document nectar
secretion through anthesis. Twelve hours after initial bag- Floral volatiles were collected in situ during daytime (06:00–
ging, the presence of nectar, its volume, and its concentra- 08:00) and nighttime (18:00–21:00) using dynamic head-
tion were recorded. Nectar was extracted with 10-mL micro- space methods (Dötterl et al. 2005; total n p 113). Individual
capillaries, and its concentration was measured using an anthetic flowers (age and pollination status not considered)
Eclipse Refractometer 45-81 (Bellingham and Stanley). Its were enclosed in polyester oven bags (10 cm#15 cm; Top-
volume was estimated from the number of filled 10-mL cap- pits, North Rhine-Westphalia, Germany), and volatiles were
illaries per flower. A subset of flowers was rebagged to assess collected for 10–30 min (depending on the strength of the
nectar replenishment at 12-h intervals (table A6). For M. perceived scent) through small adsorbent tubes (Varian Chro-
 Mixed Pollination Systems in Meriania 109

matoProbe quartz microvials; length: 15 mm, inner diameter: flowerpiercers [diurnal] and rodents [nocturnal] in Meria-
2 mm) using a membrane pump (G12/01 EB; Rietschle nia furvanthera; videos A2, A3). All flower visitors were
Thomas, Puchheim, Germany; flow rate: 200 mL/min). The foraging for nectar, which was taken up by inserting the
tubes contained 1.5 mg Tenax TA (60–80 mesh) and 1.5 mg head into the flower, thereby touching the thecae and ac-
Carbotrap B (20–40 mesh; both Supelco) fixed by glass wool tivating the saltshaker mechanism of pollen release. While
plugs (Mitchell et al. 2015). In addition, three leaf scent hummingbirds and bats mostly hovered, flowerpiercers
samples were collected at approximately 5 m distance from (passerine birds) and rodents perched. Rodents were observed
the flowers for each species (negative controls). Trapped running along branches and spent up to 10 s on a single flower
volatiles were analyzed by gas chromatography/mass spec- to drink nectar. Wasps and lepidopterans were seen as occa-
trometry (GC/MS) using an automatic thermal desorption sional nectar robbers in all species. On only a single sunny
(TD) system (TD-20; Shimadzu, Kyoto, Japan) coupled to day, small bees were observed robbing pollen in M. sanguinea.
a Shimadzu GC/MS-QP2010 Ultra equipped with a ZB-5 The insects’ contribution to pollination is likely negligible,
fused silica column (5% phenyl polysiloxane; 60 m, i.d.: as they either could not activate the saltshaker mechanism
0.25 mm, film thickness: 0.25 mm; Phenomenex). Samples (wasps, lepidopterans) or did not touch the stigmas due to
were run with a split ratio of 1∶1 and a consistent helium car- their small body size (bees). From here onward, the different
rier gas flow of 1.5 mL/min. GC oven temperature started at pollinator assemblages are grouped as follows: humming-
407C, followed by an increase of 67C/min to 2507C (held bird/bat (HB), hummingbird/rodent (HR), and flowerpiercer/
for 1 min). The MS interface worked at 2507C. Mass spectra rodent (FR).
were taken at 70 eV (electron ionization mode) from m/z 30 Visitation rates between diurnal and nocturnal func-
to 350. GC/MS data were processed using the GCMSolution tional pollinator groups differed considerably in all species,
package, version 4.11 (Shimadzu). Compound identification with higher diurnal visitation rates in M. aff. sanguinea, M.
was carried out using the ADAMS, ESSENTIALOILS-23P, tomentosa (both HB), and M. sanguinea (HR; table 1). In all
FFNSC 2, and W9N11 databases, as well as a database gener- species, both diurnal and nocturnal visitors occasionally
ated from synthetic standards available at the Plant Ecology visited more than one flower if multiple flowers were pres-
Lab at the University of Salzburg. Only compounds not pres- ent (table A3).
ent in the negative controls (i.e., flower-specific compounds)
were included in further analyses. For quantitative analysis
of volatile organic compounds, known amounts of monoter-
Pollinator Efficiency (Quality)
penes and aliphatic and aromatic compounds were injected
into the GC/MS system and mean peak areas were used to de- There were no significant differences in pollen deposition
termine the total amount of scent (Etl et al. 2016). Mann- efficiency between diurnal and nocturnal functional polli-
Whitney U-tests were used to test for significant differences nator groups in M. tomentosa (HB: t p 0:716, df p 27,
in scent release between day and night for each species sep- P p :48) and M. sanguinea (HR: t p 2 0:343, df p 14,
arately. As for nectar composition, Bray-Curtis dissimilarities P p :737), but nocturnal stigmatic pollen loads were higher
were calculated on the relative amounts of compounds and in M. aff. sanguinea (HB: t p 3:038, df p 11, P p :01).
two-way crossed PERMANOVAs run with species and day- Excluding either diurnal or nocturnal visitors did not sig-
time as factors. Relative scent compositions were visualized nificantly reduce total pollen loads compared to controls in
by nonmetric multidimensional scaling (vegan) and stacked M. tomentosa (HB) and M. sanguinea (HR) but did so in
barplots. All data have been deposited in the Dryad Digital Re- day samples of M. aff. sanguinea (HB; fig. 2; table A5).
pository: https://dx.doi.org/10.5061/dryad.jk673fq (Dellinger
et al. 2019b).
Nectar Secretion: Location
Stamens were detected as nectar-secreting organs in all
species. The exact location of nectar secretion differed be-
Results tween species, and three main types were distinguished:
secretion by dorsal filament ruptures along the entire length
Visitor Assemblages and Visitation Rates (Quantity)
of the filament (type a; figs. 1E, 1F, A1A, A1B; figs. A1–A4
Each Meriania species was visited by one diurnally active are available online), secretion by small ruptures at the ven-
functional pollinator group and a nocturnally active one tral side of the joint between filament and anther connective
(table 1; hummingbirds [diurnal] and bats [nocturnal] in (type b; fig. A1E–A1H), and secretion by porous tissue on
Meriania aff. sanguinea and Meriania tomentosa; hum- the proximal lateral sides of the filament (type c; figs. 1G,
mingbirds [diurnal] and rodents [nocturnal] in Meriania A1C, A1D). Accordingly, nectar droplets were found oozing
sanguinea; video A1 [videos A1–A3 are available online]; out of dorsal filament ruptures (visible as dark necrotic cav-
110 The American Naturalist

stamens and petals and was freely accessible to all functional

pollinator groups (fig. 1I, 1J).

Nectar Secretion: Timing and Volume

Nectar secretion started within the first 6 h of anthesis in
M. tomentosa (HB), while it started only after approxi-
mately 24 h in M. sanguinea (HR) and M. aff. sanguinea
(HB; fig. A2). Nectar was secreted throughout anthesis from
the first secretion onward and was replenished after re-
moval. In all species, pollinators started visiting flowers at
the beginning of anthesis even if there was no nectar present
yet. Nectar volume was not significantly different between
day and night (GLMM: M. tomentosa [HB]: t p 21:82,
df p 31, P p :08; M. sanguinea [HR]: t p 20:52,
df p 28, P p :61; table 2).

Nectar Concentration and Sugar Composition

Figure 2: Boxplot showing pollen deposition loads on stigmas of pol- Nectar sugar concentration ranged between 10.97 and 13.67
linator exclusion experiments in Meriania: open flower access (c; light Brix in M. aff. sanguinea, M. tomentosa (both HB), and M.
gray), day access only (D; white), and night access only (N; dark gray).
sanguinea (HR), while it was significantly higher (up to 207
There are no significant differences in Meriania tomentosa (humming-
bird/bat [HB]) and Meriania sanguinea (hummingbird/rodent); con- Brix) in M. furvanthera (FR; tables 2, A7). Only M. sanguinea
trol and night pollen loads are significantly higher in Meriania aff. san- showed significant differences in nectar concentration be-
guinea (HB). tween day and night (GLMM: M. sanguinea [HR]: t p 3:56,
df p 17, P ! :01).
Sugar composition differed significantly among species
ities) in M. tomentosa (HB; fig. 1E, 1F; type a) but sitting at (F p 114, df p 3, r2 p 0:787, P p :001; table A8), with
the filament-connective joint/upper part of the filament in sucrose being predominant in M. tomentosa, M. aff. sangui-
M. furvanthera (FR; fig. 1G; types b and c), M. aff. sanguinea nea (both HB), and M. sanguinea (HR), while hexoses were
(HB), and M. sanguinea (HR; type b). These patterns were dominant in nectar of M. furvanthera (FR; fig. 3). Meriania
confirmed by the study of six additional nectar-secreting furvanthera differed significantly from all other species (ta-
Meriania species (table A2). Ruptures of type a and b formed ble A9). Nectar sugar composition did not differ between
only in the first hours of anthesis, while porous tissue of day and night in any species or the interaction of species
type c was already present in preanthetic flowers. Regardless and day/night (table A8). Variability of nectar composition
of the exact site of secretion, nectar pooled between the differed significantly between species (F p 6:53, df p 2,

Table 2: Nectar volume, sugar content, and mean relative sugar proportions in day and night samples of the four Meriania species
Relative Relative Relative
Mean nectar amount amount amount
volume (mL) Mean 7Brix G (%) F (%) S (%) S/(F1G) F/G
Species N D N D N D N D N D N D N D
Meriania aff. sanguinea
(HB) 59.6 ... 11.8 13.1 .8 1.0 .5 1.0 98.7 98.0 71.9 53.9 .7 1.5
Meriania tomentosa (HB) 124.5 82.7 12.2 12.43 10.7 5.2 9.6 7.8 79.7 87.0 35.5 38.3 1.3 1.5
Meriania sanguinea (HR) 73.7 46.8 10.9 13.6 1.6 .7 2.1 2.0 96.3 97.3 55.5 52.1 2.1 4.1
Meriania furvanthera
(FR) ... ... 20 18.3 43.7 37.0 52.2 60.4 4.1 2.6 .04 .02 1.2 1.7
Note: Details on sample sizes are given in table A6, available online. D p measured after day; F p fructose; G p glucose; N p measured after night; S p
sucrose. Pollinator groups: FR p flowerpiercer/rodent; HB p hummingbird/bat; HR p hummingbird/rodent.
 Mixed Pollination Systems in Meriania 111

Figure 3: Triangle plot showing relative nectar sugar composition of day nectar (unfilled symbols) and night nectar (filled symbols) in the
four Meriania species. Note the clear separation following bird pollinator preferences: sucrose prevalence in hummingbird-pollinated
Meriania sanguinea, Meriania aff. sanguinea, and Meriania tomentosa and hexose dominance in flowerpiercer-pollinated Meriania fur-
vanthera. Frc p fructose; Glc p glucose; Suc p sucrose. Pollinator groups: FR p flowerpiercer/rodent; HB p hummingbird/bat; HR p hum-
mingbird/rodent.

P ! :01; figs. 3, A3) and was significantly higher in M. tomen- out as differing significantly from M. tomentosa (table A13).
tosa (HB) than in M. sanguinea (HR; table A10). Scent samples of M. sanguinea (HR) contained aliphatic com-
pounds only, with most diurnal and all nocturnal samples
containing only 1-hexen-3-one. This compound was not
Scent Composition
detected in any other species. Scents of M. furvanthera (FR)
Flowers of M. sanguinea (HR) released a strong solvent- also contained aliphatics, while scents of M. tomentosa and
like odor, and flowers of M. tomentosa (HB) produced M. aff. sanguinea (both HB) also contained terpenoids such
weak flowery odors at all times. No odors detectable by as sabinene and delta-3-carene and unknown compounds
the human nose were noted on flowers of M. aff. sanguinea (figs. 4, A4).
(HB) and M. furvanthera (FR). The GC/MS analyses re-
vealed flower-specific scent compounds in all species, how-
ever. Surprisingly, scent compounds were detected in only
Discussion
half or less of the samples in each species (table 3). Meriania
furvanthera was the only species to release detectable floral Our results suggest that the mixed vertebrate pollination
scent compounds only during daytime, while both day- syndrome in Merianieae comprises multiple bimodal polli-
and nighttime samples of the other three species contained nation systems where different functional pollinator groups
flower-specific compounds (table 3). Median total amounts can act as equally effective pollinators. These systems over-
of scent per flower per hour were significantly higher in lap in their main traits—for example, often reddish flow-
daytime samples of M. tomentosa (W p 110, df p 20, ers, day and night availability of nectar, easy reward access
P ! :01), while differences were not significant in other spe- by widely open pseudocampanulate corollas, staminal nec-
cies (table A11). Scent profiles were significantly different tar release and nectar aggregation beneath the stamens,
between species (F p 10:8, df p 3, P ! :001; table A12). and a common pollen expulsion mechanism (Dellinger
Meriania tomentosa (HB) was the only species where day and et al. 2019a). On a finer scale, certain differences in adap-
night scents differed significantly; M. sanguinea (HR) stood tation to the distinct functional pollinator groups become
Table 3: Scent composition of the four Meriania species for day (D) and night (N)
Meriania
Meriania aff. sanguinea (HB) Meriania tomentosa (HB) Meriania sanguinea (HR) furvanthera (FR)
D N D N D N D
Component RI (n p 3/6) (n p 2/4) (n p 12/25) (n p 12/23) (n p 10/23) (n p 9/24) (n p 4/8)
Aliphatic components:
methyl butanoatea 721 .0 (.0–.6)
1-hexen-3-onea 775 100.0 (.0–100.0) 100.0
3-hexanonea 785 .0 (.0–78.4)
1-hexanol 866 .0 (.0–30.9) .0 (.0–100.0) .0 (.0–63.7)
pentanoic acid 869 .0 (.0–100.0) .0 (.0–34.2)
3-octanol 995 2.1 (.0–36.1)
1-octanol 1,069 .0 (.0–15.4)
octyl acetatea 1,210 .0 (.0–42.3) 35.0 (10.2–59.7) .0 (.0–36.1)
N-bearing compounds:
2-methylbutanenitrile 723 .0 (.0–10.3)
Terpenoids:
sabinenea 979 57.7 (.0–90.2) 21.3 (13.6–29)
d-2-carene 1,006 .0 (.0–100.0)
d-3-carene 1,017 54.6 (.0–100.0)
(Z)-4,8-dimethyl-1,3,
7-nonatriene 1,099 .0 (.0–100.0)
(E)-4,8-dimethyl-1,3,
7-nonatriene 1,119 81.6 (.0–100.0)
p-1,3,8-menthatriene 1,134 .0 (.0–100.0)
b-elemene 1,408 .0 (.0–90.7)
(E)-b-caryophyllenea 1,437 .0 (.0–100.0) 15.0 (3.0–26.9) .0 (.0–69.1) .0 (.0–55.4)
(E)-b-farnesenea 1,463 1.2 (.0–2.5)
a-humulenea 1,481 5.8 (.1–11.6)
Unknowns:
m/z 40,41,55,69,107,119 1,084 .0 (.0–1.4)
m/z 55,69,81,109,135,161 1,487 .0 (.0–100.0)
m/z 40,41,69,94,120,135 1,493 8.7 (4.4–12.9)
m/z 40,55,69,135,187,223 1,505 1.0 (.4–1.6)
m/z 40,69,105,121,136,177 1,508 1.6 (.0–2.1)
m/z 41,55,69,102,121,136 1,518 .9 (.0–18.6)
Total amount of scent
ng/flower/h 4.8 (2.8–8.5) 17.6 (16.3–18.9) 21.3 (2.8–78.7) 4.1 (.1–15.0) 31.9 (.9–184.1) 31.3 (2.0–87.4) 24.4 (2.2–58.6)
Note: n indicates the number of samples that contained detectable scent compounds out of the total number of samples analyzed. Median relative amounts of the specific volatile compounds are given, and
compounds are arranged according to chemical class and sorted by the Kovats retention index (RI; ZB-5 column). Minimum/maximum values are given in parentheses. Total median amount of scent released per
flower per hour (ng) is given in the final row. For M. furvanthera, only diurnal scent samples contained compounds, while zero of the two analyzed night scent samples contained compounds. Pollinator groups: FR p
flowerpiercer/rodent; HB p hummingbird/bat; HR p hummingbird/rodent.
a
Identified based on the retention time and mass spectra of synthetic standard compounds.
 Mixed Pollination Systems in Meriania 113

Figure 4: Nonmetric multidimensional scaling based on a Bray-Curtis dissimilarity matrix to display semiquantitative differences in day and
night scent profiles of the four Meriania species. The stress value of 0.018 indicates a good representation of the observed similarities among
scent samples. The six compounds correlating best with the coordinates are given. Pollinator groups: FR p flowerpiercer/rodent; HB p
hummingbird/bat; HR p hummingbird/rodent.

apparent: nectar sugar composition follows typical diurnal ther pollinator group did not significantly reduce stigma
bird pollinator preferences (Johnson and Nicolson 2008), pollen loads as compared to open controls in these two spe-
and scent profiles partially show adaptations to the differ- cies. This merits further investigation, as it could indicate
ent nocturnal pollinators. that each plant species could successfully reproduce if vis-
Our finding of effective rodent pollination in Meriania ited by one pollinator group only. In M. aff. sanguinea, bats
sanguinea and Meriania furvanthera is particularly inter- were more effective pollinators than hummingbirds. How-
esting given the rarity of documented cases of rodent pol- ever, there are clearly more aspects to pollinator quality than
lination in general and especially in the New World (e.g., only pollen deposition (but see Muchhala et al. 2009 for a
Melastomataceae [Lumer 1980]; Loasaceae [Cocucci and similar approach to ours). Quality differences between polli-
Sérsic 1998]; Proteaceae [Cárdenas et al. 2017]). Both spe- nators also involve differences in the efficiency of removing
cies with rodent pollination show modifications in their pollen, which then gets deposited (and not lost), and the
inflorescence architecture (short-pediceled flowers in leaf “purity” of deposited pollen (e.g., amount of heterospecific
axils in M. furvanthera; fig. 1B) or growth form (procum- pollen; see Morales and Traveset 2008; Queiroz et al. 2015),
bent habit of M. sanguinea; fig. 1C), which facilitate access as well as the genetic compatibility/viability of deposited pol-
to flowers by perching pollinators. This is in contrast to len (e.g., self-/outcross pollen and consequently fitness of
flowers protruding on long inflorescence stalks in Meriania offspring; Minnaar et al. 2018). Manual pollination experi-
tomentosa and Meriania aff. sanguinea (fig. 1A), which are ments in M. sanguinea and M. tomentosa showed self-
visited only by pollinators capable of hovering while drink- compatibility (A. S. Dellinger, unpublished data). Thus, more
ing nectar (HB). Although rodent visitation rates were fine-grained assessments of stigmatic pollen loads could
10 times lower than hummingbird visitation rates in M. san- bring out subtle quality differences between the different
guinea (table 1), rodents contributed substantially to pollen pollinator groups in the future.
deposition on stigmas and hence must be considered as le- We found nectar secretion only from stamens, which
gitimate pollinators (fig. 2). Likewise, hummingbirds were contradicts findings on hypanthial nectar secretion in Meria-
more frequent visitors than bats in M. tomentosa and M. nieae (Varassin et al. 2008; but see also Stein and Tobe 1989).
aff. sanguinea but deposited the same or lower amounts of Although the exact location of nectar secretion is variable,
pollen. It is possible that the relatively small experimental the systems are overall similar in having unspecialized sta-
sample sizes have reduced the power of detecting significant minal nectaries with direct connection to the phloem. Pos-
differences between the diurnal and nocturnal pollinators in sibly, the pronounced stamen movement in early stages of
M. sanguinea and M. tomentosa. Interestingly, excluding ei- anthesis (fig. 1H–1J) leads to high pressure in the tissue,
114 The American Naturalist

which causes tissue rupture and phloem sap leakage (Vogel tracts its mammal pollinators. Interestingly, these results
1997; De la Barrera and Nobel 2004). This idea is supported are in line with a study reporting lack of floral scent in other
by our finding that ruptures form only within the first hours Melastomataceae species (genus Blakea) for which rodent
of anthesis but are not present in the bud stage. Generally, visitation has been reported (Lumer 1980; Wester et al.
invertases can change sucrose-rich phloem composition in 2016). Furthermore, it is notable that all four Meriania taxa
the nectary (Nicolson et al. 2007), and plants have been found released scents during daytime (table 3). In traditional polli-
to even be capable of changing their nectar composition be- nation syndrome theory, “bird” flowers usually attract pol-
tween day and night (e.g., in Inga sessilis; Amorim et al. linators by visual cues (showy red flowers) rather than scent
2013). In the Meriania species studied here, nectar sugar (Dobson et al. 2006). More recent studies, however, indicate
composition did not change between day and night and that birds use olfactory cues in addition to vision when for-
sugar compositions corresponded to preferences described aging (Kessler and Baldwin 2007).
for bird pollinators (Johnson and Nicolson 2008). We found Taken together, our results support the view that nectar-
a clear differentiation between specialized nectar feeders producing Meriania species, summarized into a mixed ver-
(hummingbirds, sucrose rich: M. tomentosa, M. aff. sanguinea, tebrate pollination syndrome, comprise different bimodal
M. sanguinea) and more generalist nectar feeders (flower- pollination systems. While studies on nectar-secreting Mela-
piercers, hexose rich: M. furvanthera; fig. 3). The hexose-rich stomataceae from other tribes (e.g., Miconieae) report an
nectar of M. furvanthera, however, indicates the presence increased “generalization” (e.g., Kriebel and Zumbado 2014;
of nectary invertases despite the unspecialized nectar leak- Brito et al. 2017), our mixed vertebrate syndrome is better
age (De la Barrera and Nobel 2004; for hexose-rich food described as “specialized bimodal” (cf. Manning and Gold-
bodies in closely related passerine-pollinated Axinaea, see blatt 2005). Such bimodal systems have been considered
also Dellinger et al. 2014). The origin of the large variability as labile, possibly representing evolutionary transitions be-
in nectar sugar composition in M. tomentosa remains un- tween distinct pollination syndromes (Manning and Gold-
known but could be interpreted as a means of meeting both blatt 2005). Given the ancestral bee-buzz pollination syn-
hummingbird and bat preferences (Abrahamczyk et al. drome in Merianieae, one could expect such transitions
2017). between (ancestral) bee pollinators and a (derived) verte-
Contrary to our expectation of increased floral scent re- brate pollinator or further transitions between two func-
lease during nighttime as an adaptation to bat and rodent at- tional vertebrate pollinators (e.g., hummingbird to bat;
traction (Dobson 2006), flowers did not smell more strongly Rosas-Guerrero et al. 2014). Alternatively, bimodal polli-
at night. At the level of scent classes, M. tomentosa and M. nation systems in Meriania could have arisen without prior
aff. sanguinea (both HB) released higher amounts of ter- specialization on one new functional group but actually rep-
penoids, known to be important in bat pollination, while resent stable systems adapted to exploit two complementary
aliphatic compounds were dominant in rodent-pollinated groups of pollinators. This scenario seems plausible in
M. sanguinea and M. furvanthera (fig. A4; Knudsen et al. Meriania given the lack of (ancestral) bee pollinators in the
1995; Pettersson et al. 2004; Dobson et al. 2006). Meriania mixed vertebrate syndrome and the fact that there is, to
sanguinea is particularly interesting in this context: 1- date, no nectar-secreting Meriania species known to be pol-
hexen-3-one (mostly confined to nocturnal scent samples) linated by only one type of vertebrate pollinator (either hum-
is known as flower scent only from Cytinus visseri (Cytina- mingbirds, flowerpiercers, bats, or rodents). The repeated
ceae, Malvales), a parasitic South African plant pollinated independent origin of different bimodal systems (shift 1:
by rodents and shrews (Johnson et al. 2010). Curiously, 1- M. tomentosa [HB], M. furvanthera [FR]; shift 2: M. aff.
hexen-3-one worked as a repellent when tested alone in a sanguinea [HB], M. sanguinea [HR]) further supports the
pollinator behavioral assay but had no negative effects idea of a stable pollination strategy. This is particularly in-
when tested in combination with the strong attractant 3- teresting given that much work on pollinator shifts has
hexanone, also released by C. visseri (Johnson et al. 2010). found strong adaptive trade-offs in species visited by two
In M. sanguinea, however, 3-hexanone was detected only functional pollinator groups (Castellanos et al. 2004; Much-
during daytime when rodents are not active. Thus, the role hala 2007). While clear “pro-bird, anti-bee” changes in floral
of 1-hexen-3-one as pollinator attractant in M. sanguinea traits have been documented in Penstemon (Castellanos et al.
remains equivocal. At the larger scale, however, the simulta- 2004), our results indicate that Meriania flowers can success-
neous occurrence of 1-hexen-3-one in plants of different fully exploit two different functional pollinator groups. Tak-
orders (Myrtales, Malvales) and continents (South America, ing this thought further, our results suggest that certain com-
Africa) points toward convergence in the evolution of this binations of pollinators, such as hummingbirds and bats,
compound to communicate with ground-dwelling mam- may potentially come without (or with very small) fitness
mals. Given the lack of detectable scent compounds at night trade-offs, at least in some systems. More detailed experi-
in M. furvanthera, it remains unclear how this species at- ments are needed, however, to clarify these ideas.
 Mixed Pollination Systems in Meriania 115

Acknowledgments De la Barrera, E., and P. S. Nobel. 2004. Nectar: properties, floral as-
pects, and speculations on origin. Trends in Plant Science 9:65–69.
We thank the Bellavista Rainforest Lodge, the Jatun Sacha Dellinger, A. S., M. Chartier, D. Fernández-Fernández, D. S.
Foundation, the Jocotoco Foundation, and staff at Caja- Penneys, M. Alvear, F. Almeda, F. A. Michelangeli, Y. Staedler,
numa (Podocarpus National Park) in Ecuador and Estación W. S. Armbruster, and J. Schönenberger. 2019a. Beyond buzz-
Biológica Monteverde in Costa Rica for providing housing pollination—departures from an adaptive plateau lead to new
pollination syndromes. New Phytologist 221:1136–1149.
at field sites. We further thank Hanna Schneeweiss for pro-
Dellinger, A. S., D. S. Penneys, Y. M. Staedler, L. Fragner, W.
viding access to a fluorescence microscope, Jorge Brito for Weckwerth, and J. Schönenberger. 2014. A specialized bird polli-
confirming rodent identifications, and Nathan Muchhala nation system with a bellows mechanism for pollen transfer and
for confirming bat identifications. This study was supported staminal food body rewards. Current Biology 24:1615–1619.
by Austrian Science Fund grant P-30669 to A.S.D. and J.S. Dellinger, A. S., L. M. Scheer, S. Artuso, D. Fernández-Fernández,
and National Science Foundation Division of Environmen- F. Sornoza, D. S. Penneys, R. Tenhaken, S. Dötterl, and J. Schönen-
tal Biology grant 1543721 to D.S.P. berger. 2019b. Data from: Bimodal pollination systems in An-
dean Melastomataceae involving birds, bats, and rodents. Ameri-
can Naturalist, Dryad Digital Repository, https://dx.doi.org/10.5061
/dryad.jk673fq.
Literature Cited Dobson, H. E. M. 2006. Relationship between floral fragrance com-
position and type of pollinator. Pages 147–198 in N. Dudareva and
Abrahamczyk, S., M. Kessler, D. Hanley, D. N. Karger, M. P. J. E. Pichersky, eds. Biology of floral scent. CRC, Boca Raton, FL.
Müller, A. C. Knauer, F. Keller, M. Schwerdtfeger, and A. M. Hum- Dötterl, S., L. M. Wolfe, and A. Jürgens. 2005. Qualitative and quantitative
phreys. 2017. Pollinator adaptation and the evolution of floral nectar analyses of flower scent in Silene latifolia. Phytochemistry 66:203–213.
sugar composition. Journal of Evolutionary Biology 30:112–127. Dupont, Y. L., D. M. Hansen, J. T. Rasmussen, and J. M. Olesen.
Aigner, P. A. 2001. Optimality modelling and fitness trade-offs: when 2004. Evolutionary changes in nectar sugar composition associated
should plants become pollinator specialists? Oikos 95:177–184. with switches between bird and insect pollination: the Canarian
———. 2006. The evolution of specialized floral phenotypes in a bird-flower element revisited. Functional Ecology 18:670–676.
fine-grained environment. Pages 23–46 in N. M. Waser and J. Etl, F., A. Berger, A. Weber, J. Schönenberger, and S. Dötterl. 2016. Noctur-
Ollerton, eds. Plant-pollinator interactions: from specialization to nal plant bugs use cis-jasmone to locate inflorescences of an Araceae
generalization. University of Chicago Press, Chicago. as feeding and mating site. Journal of Chemical Ecology 42:300–304.
Akter, A., P. Biella, and J. Klecka. 2017. Effects of small-scale cluster- Faegri, K., and L. van der Pijl. 1979. The principles of pollination
ing of flowers on pollinator foraging behaviour and flower visita- ecology. Pergamon, Oxford.
tion rate. PLoS ONE 12:e0187976. Fenster, C. B., W. S. Armbruster, P. Wilson, M. R. Dudash, and J. D.
Amorim, F. W., L. Galetto, and M. Sazima. 2013. Beyond the polli- Thomson. 2004. Pollination syndromes and floral specialization.
nation syndrome: nectar ecology and the role of diurnal and noctur- Annual Review of Ecology, Evolution, and Systematics 35:375–403.
nal pollinators in the reproductive success of Inga sessilis (Fabaceae). Fleming, T. H., C. Geiselman, and W. J. Kress. 2009. The evolution
Plant Biology 15:317–327. of bat pollination: a phylogenetic perspective. Annals of Botany
Baker, H. G., and I. Baker. 1983. Floral nectar sugar constituents in 104:1017–1043.
relation to pollinator type. Pages 126–152 in C. E. Jones and R. J. Johnson, S. D., and S. W. Nicolson. 2008. Evolutionary associations
Little, eds. Handbook of experimental pollination biology. Van between nectar properties and specificity in bird pollination sys-
Nostrand Reinold, New York. tems. Biology Letters 4:49–52.
Baker, H. G., I. Baker, and S. A. Hodges. 1998. Sugar composition of Johnson, S. D., P. M. Burgoyne, L. D. Harder, and S. Dötterl. 2010.
nectars and fruits consumed by birds and bats in the tropics and Mammal pollinators lured by the scent of a parasitic plant. Pro-
subtropics. Biotropica 30:559–586. ceedings of the Royal Society B 278:2303–2310.
Brito, V. L. G., A. R. Rech, J. Ollerton, and M. Sazima. 2017. Nectar Kearns, C. A., and D. W. Inouye. 1993. Techniques for pollination
production, reproductive success and the evolution of generalised biologists. University Press of Colorado, Boulder.
pollination within a specialised pollen-rewarding plant family: a Kessler, D., and I. T. Baldwin. 2007. Making sense of nectar scents:
case study using Miconia theizans. Plant Systematics and Evolu- the effects of nectar secondary metabolites on floral visitors of Ni-
tion 303:709–718. cotiana attenuata. Plant Journal 49:840–854.
Buchmann, S. L. 1983. Buzz pollination in angiosperms. Pages 73– Knudsen, J. T., R. Eriksson, J. Gershenzon, and B. Ståhl. 2006. Diver-
113 in C. E. Jones and R. J. Little, eds. Handbook of experimental sity and distribution of floral scent. Botanical Review 72:1–120.
pollination biology. Van Nostrand Reinold, New York. Knudsen, J. T., and L. Tollsten. 1995. Floral scent in bat-pollinated
Cárdenas, S., C. Niuvelo-Villavicencio, J. D. Cárdenas, O. P. Landázuri, plants: a case of convergent evolution. Botanical Journal of the
and B. A. Tinoco. 2017. First record of flower visitation by a rodent Linnean Society 119:45–57.
in Neotropical Proteaceae, Oreocallis grandiflora. Journal of Tropical Kriebel, R., and M. A. Zumbado. 2014. New reports of generalist insect
Ecology 33:174–177. visitation to flowers of species of Miconia (Miconieae: Melastoma-
Castellanos, M. C., P. Wilson, and J. D. Thomson. 2004. “Anti-bee” taceae) and their evolutionary implications. Brittonia 66:396–404.
and “pro-bird” changes during the evolution of hummingbird pollina- Kuznetsova, A., P. B. Brockhoff, and R. H. B. Christensen. 2017.
tion in Penstemon flowers. Journal of Evolutionary Biology 17:876–885. lmerTest: tests in linear mixed effects models. R package version 3.0-1.
Cocucci, A. A., and A. N. Sérsic. 1998. Evidence of rodent pollina- https://cran.r-project.org/web/packages/lmerTest/index.html.
tion in Cajophora coronata (Loasaceae). Plant Systematics and Evo- Lumer, C. 1980. Rodent pollination of Blakea (Melastomataceae) in
lution 211:113–128. a Costa Rican cloud forest. Brittonia 32:512–517.
116 The American Naturalist

Manning, J. C., and P. Goldblatt. 2005. Radiation of pollination sys- Raguso, R. A., R. A. Levin, S. E. Foose, M. W. Holmberg, and
tems in the Cape genus Tritoniopsis (Iridaceae: Crocoideae) and L. A. McDade. 2003. Fragrance chemistry, nocturnal rhythms
the development of bimodal pollination strategies. International and pollination “syndromes” in Nicotiana. Phytochemistry 63:265–
Journal of Plant Sciences 166:459–474. 284.
Martinez Arbizu, P. 2017. pairwiseAdonis: pairwise multilevel com- Renner, S. S. 1989. A survey of reproductive biology in Neotropical
parison using adonis. R package version 0.0.1. Melastomataceae and Memecylaceae. Annals of the Missouri Bo-
Minnaar, C., B. Ancerson, M. L. de Jager, and J. D. Karron. 2018. tanical Garden 50:496–518.
Plant-pollinator interactions along the pathway to paternity. An- Ridgely, R. S., and P. J. Greenfield. 2001. The birds of Ecuador: field
nals of Botany 123:225–245. guide. Cornell University Press, Ithaca, NY.
Mitchell, T. C., S. Dötterl, and H. Schaefer. 2015. Hawk-moth polli- Rosas-Guerrero, V., R. Aguilar, S. Martén-Rodríguez, L. Ashworth,
nation and elaborate petals in Cucurbitaceae: the case of the Ca- M. Lopezaraiza-Mikel, J. M. Bastida, and M. Quesada. 2014. A
ribbean endemic Linnaeosicyos amara. Flora 216:50–56. quantitative review of pollination syndromes: do floral traits pre-
Morales, C. L., and A. Traveset. 2008. Interspecific pollen transfer: dict effective pollinators? Ecology Letters 17:388–400.
magnitude, prevalence and consequences for plant fitness. Critical Stebbins, G. L. 1970. Adaptive radiation of reproductive characteristics
Reviews in Plant Sciences 27:221–238. in angiosperms. I. Pollination mechanisms. Annual Review of Ecol-
Muchhala, N. 2007. Adaptive trade-off in floral morphology mediates ogy and Systematics 1:307–326.
specialization for flowers pollinated by bats and hummingbirds. Stein, B. A., and H. Tobe. 1989. Floral nectaries in Melastomataceae
American Naturalist 169:494–504. and their systematic and evolutionary implications. Annals of the
Muchhala, N., A. Caiza, J. C. Vizuete, and J. D. Thomson. 2009. A Missouri Botanical Garden 76:519–531.
generalized pollination system in the tropics: bats, birds and Thomson, J. D. 2003. When is it mutualism? American Naturalist
Aphelandra acanthus. Annals of Botany 103:1481–1487. 162(suppl.):S1–S9.
Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. Fonseca, Tirira, D. 2017. Guía de campo de los Mamíferos del Ecuador. Asocia-
and J. Kent. 2000. Biodiversity hotspots for conservation priori- ción Ecuatoriana de Mastozoología y Editorial Murciélago Blanco,
ties. Nature 403:853–858. Quito, Ecuador.
Nicolson, S. W., M. Nepi, and E. Pacini, eds. 2007. Nectaries and Varassin, I. G., D. S. Penneys, and F. A. Michelangeli. 2008. Compar-
nectar. Springer, Dordrecht. ative anatomy and morphology of nectar-producing Melastomata-
Oksanen, J., F. G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. ceae. Annals of Botany 102:899–909.
McGlinn, P. R. Minchin, et al. 2018. vegan: community ecology Vogel, S. 1997. Remarkable nectaries: structure, ecology, organo-
package. R package version 2.5-2. https://cran.r-project.org/web phyletic perspectives. I. Substitutive nectaries. Flora 192:305–333.
/packages/vegan/index.html. Waser, N. M., L. Chittka, M. V. Price, N. M. Williams, and J.
Pettersson, S., F. Eryk, and J. T. Knudsen. 2004. Floral scent of bat- Ollerton. 1996. Generalization in pollination systems, and why it
pollinated species: West Africa vs. the New World. Biological matters. Ecology 77:1043–1060.
Journal of the Linnean Society 82:161–168. Wester, P., M. Filla, and K. Lunau. 2016. Floral scent and flower
Queiroz, J. A., Z. G. M. Quirino, A. V. Lopes, and I. C. Machado. 2016. visitors of three green-flowered Costa Rican and Panamanian Blakea
Vertebrate mixed pollination system in Encholirium spectabile: a species (Melastomataceae) indicate birds rather than rodents as
bromeliad pollinated by bats, opossum and hummingbirds in a pollinators. Plant Ecology and Evolution 149:319–328.
tropical dry forest. Journal of Arid Environments 125:21–30. Wurdack, J. J. 1967. Meriania sanguinea. Memoirs of the New York
Queiroz, J. A., Z. G. M. Quirino, and I. C. Machado. 2015. Floral Botanical Garden 16:4.
traits driving reproductive isolation of two co-flowering taxa that
share vertebrate pollinators. Annals of Botany PLANTS 7:plv127. Natural History Editor: Edmund D. Brodie III

Black-thighed puffleg (Eriocnemis derbyi) sitting on infructescence of Meriania tetragona after visiting its flowers in the Tapichalaca Reserve
(Jocotoco Foundation). Photo q Francisco Sornoza.
Copyright of American Naturalist is the property of University of Chicago and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.