Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2024 Dec 6;135(4):669–680. doi: 10.1093/aob/mcae210

A modified petal and stamen dimorphism interact to enhance pollen placement by a buzz-pollinated flower

Thainã R Monteiro 1,✉, Rogério V S Gonçalves 2, Francismeire J Telles 3, Gudryan J Barônio 4, Anselmo Nogueira 5, Vinícius L G Brito 6
PMCID: PMC11904892  PMID: 39657108

Abstract

Background

Floral adaptations supposedly help pollen grains to cross the numerous barriers faced during their journey to stigmas. Stamen dimorphism and specialized petals, like the cucculus in the Cassieae tribe (Fabaceae), are commonly observed in flowers that offer only pollen as a resource for bee pollinators. Here, we experimentally investigated whether stamen dimorphism and the cucculus enhance pollen placement on the bee’s body.

Methods

We used 3-D-printed bee models to apply artificial vibrations to the flowers of Chamaechrista latistipula with their cucculus deflected or maintained in its original position and their anther pores manipulated. After each simulated flower visit, we captured photographs of the artificial bee from four distinct angles. Employing digital imaging techniques, we documented the presence and location of pollen and stigma on the bee’s body.

Key Results

Our findings reveal that the cucculus redistributes pollen grains on the bee’s body. There is a remarkable increase in pollen density (~10-fold) on the lateral side adjacent to the cucculus, precisely where the stigma contacts the bee when the cucculus is unmanipulated. Furthermore, the cucculus also enhances pollen placement on the ventral region of the bee, indicating its additional function. The cucculus also increases the accuracy of pollen grains on the adjacent lateral region of the bee’s body, irrespective of the pollen grains released by small or large anthers.

Conclusions

Floral specialized traits, such as modified petals and stamen dimorphism, can modify the fate of pollen grains and ultimately contribute to male reproductive performance in pollen flowers with poricidal anthers. The cucculus exhibits a dual role by promoting pollen placement in optimal regions for pollination and probably supporting pollen grains for bee feeding. These findings provide valuable insights into the adaptive significance of floral traits and their impact on the reproductive success of pollen flowers.

Keywords: Buzz pollination, floral morphology, heteranthery, inaccuracy, male performance, poricidal anthers

INTRODUCTION

One of the most notable examples of flower specialization is buzz-pollinated flowers that conceal their pollen grains within poricidal anthers (Vallejo-Marín et al., 2010; Vallejo‐Marín, 2019). These flowers are pollinated by female bees able to produce vibrations with their thoracic muscles, which in turn are mechanically transmitted to anthers, resulting in pollen release and a buzzing sound as a byproduct (Vogel, 1978; Buchmann, 1983). Buzz pollination independently evolved multiple times during angiosperm diversification, occurring in more than 22 000 extant plant species (Lunau, 2004; Vallejo-Marín and Russell, 2024; Russell et al., 2024). As most buzz-pollinated flowers offer only pollen as a resource for buzzing bees, this system is thought to be under an intense ‘pollen dilemma’ – an evolutionary challenge of balancing pollen allocation between pollinator reward and plant fertilization (Westerkamp, 1996; Westerkamp, 1997).

In addition to poricidal anthers, stamen dimorphism (i.e. more than one type of stamen within the same flower) is another common trait that emerged in buzz-pollinated flowers, potentially minimizing the pollen dilemma and increasing the accuracy of pollen transfer to the stigma (Renner, 1990; Luo et al., 2009; Vallejo-Marín et al., 2010; Cardoso et al., 2018; Konzmann et al., 2020). The functional role of stamen dimorphism (or heteranthery) was first hypothesized over a century ago (Müller, 1881, 1882, 1883), although experimental investigations into it were made only very recently (Luo et al., 2009; Vallejo-Marín et al., 2009, 2010; Paulino et al., 2013; Paulino et al., 2016; Li et al., 2015; Mesquita‐Neto et al., 2017; Saab et al., 2021). In buzz-pollinated flowers with stamen dimorphism, bees preferentially forage on the central conspicuous feeding anthers, while pollinating anthers are more external and less attractive to the bees, resulting in their pollen grains being more likely to be deposited on the stigma (Luo et al., 2008; Vallejo-Marín et al., 2009; Velloso et al., 2018; Saab et al., 2021). Both poricidal anthers and stamen dimorphism are understood to be a specialization for buzz pollination, which would increase the accuracy of pollen transfer and deposition, favouring the sexual reproduction of these plants.

Some groups of plants even have additional floral traits associated with the buzz-pollination process. A remarkable and recurrent feature of buzz-pollinated plants belonging to the tribe Cassieae (Fabaceae) (Rando et al., 2024) is the presence of modified petals associated with the poricidal anthers (Marazzi and Endress, 2008; Almeida et al., 2015a, b). In these flowers, pollen grains ejected from the poricidal anthers may have their trajectory modified by these petals during buzz pollination (Delgado-Salinas and Sousa-Sánchez, 1977; Westerkamp, 2004; Amorim et al., 2017; Nogueira et al., 2018), distributing the pollen grains on different regions of the bee’s body relative to where the poricidal anthers would directly place them. This mechanism possibly increases pollen placement accuracy, which is fundamental to male fitness during buzz pollination. Despite the widespread occurrence of these modified petals and the dimorphic poricidal anthered stamens in flowers of the Cassieae tribe, the extent of their impact on pollen placement success has never been assessed.

In this study, we experimentally investigate the role of modified petals and stamen dimorphism in pollen placement in buzz-pollinated flowers with poricidal anthers. We conducted a greenhouse experiment using the legume plant species Chamaecrista latistipula with poricidal flowers bearing different types of stamens and a curved and rigid modified petal known as a ‘cucculus’ (Nogueira et al., 2018; Rando et al., 2021). This species belongs to the Cassieae tribe, one of the largest buzz-pollinated groups, in which stamen dimorphism and cucculus are prevalent. To explore the functionality of the distinct stamens and cucculus, we manipulated the flower structure, partially or totally obstructing the anther pores of the different stamens sets and bending or not bending the cucculus. Next, we applied artificial mechanical vibrations with a bee pollinator mimic on the flowers under different conditions. Then, we quantified pollen placement and the stigma position on the bee’s body using digital imaging techniques. Within the scope of this experiment, we aim to address three key questions: (1) How does the cucculus impact the pollen deposition density in different regions of the bee’s body? (2) How do different stamens influence the pollen density on the lateral regions of the bee’s body adjacent to the cucculus? (3) To what extent do the cucculus and stamen dimorphism reduce inaccurate pollen placement on the bee’s body, and how do they contribute to male reproductive performance in these flowers? We hypothesize that the cucculus and large stamens increase pollen density and placement accuracy on the bees’ bodies, potentially increasing male reproductive performance in buzz-pollinated flowers.

MATERIALS AND METHODS

Flower species and study site

Chamaecrista latistipula (Fig. 1) is a shrubby species that exhibits wide distribution in the Brazilian Cerrado ecosystem (Nogueira et al., 2018). During its flowering season between October and March (Fidalgo et al., 2018), this plant species produces vibrant yellow flowers that attract hymenopteran and coleopteran visitors (Ruhren, 2003; Camargo and Miotto, 2004). The flowers of C. latistipula last one light day and are characterized by having five petals, with one of them forming a distinctive and specialized structure known as the cucculus. The cucculus is the innermost lateral petal, concave, slightly thicker, and more robust than other petals in C. latistipula flowers (Nogueira et al., 2018). Additionally, these flowers display stamen dimorphism, with three large stamens and seven smaller ones that are centrally positioned. Unlike other species within Cassiinae, where larger and smaller stamens occupy different positions within the flower and have different colours for bee pollinators (e.g. Saab et al., 2021), the stamens of C. latistipula are similar in colour and positioned centrally in the flower, differing only in size. Moreover, the seven smaller stamens contribute two-thirds of the pollen grains in the flower, while the remaining third is distributed among the three larger stamens (A. Nogueira, unpubl. data). Therefore, the division of labour in the C. latistipula flower is likely much less pronounced, and it becomes challenging to assess beforehand the significance of stamen size dimorphism in directing pollen to the bee’s body, given that their arrangement can be influenced by the modified petal, the cucculus (Nogueira et al., 2018).

Fig. 1.

Fig. 1.

Open in a new tab

(A) Chamaecrista latistipula flower being visited by a buzzing bee, Bombus morio. (B) Front view of C. latistipula flower with unmanipulated cucculus. (C) Front view of C. latistipula with deflected cucculus. (D) 3-D artificial bee model attached to a wooden rod connected to the vibration transductor speaker featuring a vibrating metal plate. Chamaecrista latistipula flowers are characterized by five bright yellow petals, one of them forming a distinctive curved structure known as the cucculus. Also, the flower displays stamen dimorphism, with three large stamens and seven smaller ones that are centrally positioned and bear poricidal anthers.

Chamaecrista latistipula plants also exhibit non-reciprocal monomorphic enantiostyly, producing mirror-image flowers within the same plant. Half of the flowers produced by an individual C. latistipula plant have styles deflected to the right, while the other half have styles deflected to the left. Although the positioning of anthers relative to the styles is not reciprocal, the cucculus always shows deflection in the opposite direction compared with the style within the same flower.

The experiment took place in October 2021 at the Federal University of ABC in São Bernardo do Campo, São Paulo, Brazil, coinciding with the peak of the flowering phase of C. latistipula. There, an experimental population of C. latistipula, initially transplanted from a natural population in Mairiporã, São Paulo (latitude −23.320177, longitude −46.584018), was used as a seed source. Seeds obtained from these specimens were sown in 11-L pots, containing a layer of expanded clay at the bottom and filled with a homogenized substrate comprising sand and vegetal soil. Subsequently, the seeds were planted superficially at 0.5–1 cm below the soil surface. The plants were cultivated in a greenhouse at the university and irrigated using automatic irrigators positioned near the pots for 2 years before the experiments.

Artificial vibrations

We constructed a 3-D artificial bee model to examine the impact of the cucculus and stamen dimorphism on pollen density and inaccuracy when placed on a bee’s body. Artificial bees were used to keep the bee size and vibrations standardized and to avoid killing bees. The model was created using a 3-D printer (Anycubic Photon Mono X) and black resin material (Anycubic Black, viscosity at 25 °C of 552 mPa s). Although it was impossible to print bee hairs, the pollen grains visually attached well to the resin material, making pollen placement comparable among treatments. The dimensions of the artificial bee’s body were determined based on the length and intertegular distance of Bombus morio, a key pollinator of C. latistipula flowers (Fidalgo et al., 2018). This artificial bee model was affixed to a pin, which was attached to a wooden rod measuring 6 cm in height and 4 mm in diameter. The rod and the longitudinal axis of the bee’s body formed a 45° angle. Subsequently, the wooden rod was connected to a vibration transducer speaker featuring a vibrating metal plate (Youts Globe Super Speaker). This speaker transduced signals designed to simulate the vibrations produced by bees during their flower visitation.

The artificial vibrational signals were generated on a computer using Audacity software (www.audacityteam.org). We synthesized sine waves with a frequency of 300 Hz, a relative amplitude of 1, a sampling rate of 44.1 kHz and a phase φ = 0°. These signals were designed to mimic the vibrations produced by buzzing bees during their visitation to C. latistipula flowers, drawing upon different systems described in previous studies (de Luca et al., 2013; Arroyo-Correa et al., 2019; Brito et al., 2020; Pritchard and Vallejo-Marín, 2020). The artificial vibration consisted of eleven 1-s pulses separated by 200 ms of silence. The transducer speaker’s absolute amplitude velocity was directly adjusted to achieve 100 mm s−1.

We used a laser Doppler vibrometer (PDV-100, Polytec) for system calibration, ensuring the signals reproduced as vibration in the artificial bees remained consistent during the experiment. The laser beam was directed parallel to the main displacement axis of the transducer, continuously targeting the artificial bee’s head. The vibrometer settings included a low-pass filter of 5 Hz, no high-pass filter, and a laser speed of 500 mm s−1. The calibration vibrometer was connected to a secondary computer, enabling real-time monitoring of the artificial bee’s vibration using the VibSoft-20 software in the oscilloscope function. Supplementary Data Fig. S1 shows the details of the experimental setup.

Experiment and image treatments

We conducted a 2 × 3 factorial experiment manipulating the cucculus (two levels) and anther pores (three levels) to investigate the impact of the cucculus and stamen dimorphism on pollen placement. Six different conditions were created manipulating both structures as follows: (1) flowers with unmanipulated cucculus and both types of anthers unobstructed (complete control); (2) flowers with unmanipulated cucculus and unobstructed small anthers; (3) flowers with unmanipulated cucculus and unobstructed large anthers; (4) flowers with deflected cucculus and both types of anthers unobstructed; (5) flowers with deflected cucculus and unobstructed small anthers; and (6) flowers with deflected cucculus and unobstructed large anthers. We performed 36 replicates on each treatment (see Supplementary Data Table S1 for replication statement), using a new fresh flower for each replicate. The number of replicates analysed may differ from this original number because some data were impossible to recover due to a lack of contrast between the pollen grains and bee resin, spoiling a proper binarization during image processing (see below, Supplementary Data Fig. S3).

To prevent any alteration of floral biomechanical properties resulting from mass loss or lack of contact with the artificial bee, we manipulated the cucculus by deflecting it instead of removing it from the flowers. The obstruction of large or small anthers was achieved using methacrylate superglue (Locite Super Bonder). We also placed an equivalent drop of superglue on the unobstructed anthers to control for any potential effect of this manipulation. Subsequently, the treated flowers were carefully detached from the plants and manually secured to the artificial bee coupled to the vibration transducer. Following the vibration, we marked the position of the stigmas on the side of the bee’s body opposite the cucculus using fluorescent powder (Pigmento Fluorescente em Pó Rosa, Hazzin). The artificial bee was then decoupled from the vibration transducer and photographed (using either a Canon EOS Rebel SL3 camera with a 100-mm macro lens or a Samsung NX300 camera) from four different angles, emphasizing its lateral region adjacent to the cucculus, dorsal region, lateral region opposite to the cucculus, and ventral region, while placed on a black foam stage (Supplementary Data Fig. S2). Due to the variation in sensor characteristics between the two cameras, images captured with the Samsung NX300 were resampled to match the resolution of those taken with the Canon EOS Rebel SL3. We used the nearest neighbour interpolation technique, which ensures uniform spatial characteristics across all datasets, preserving the original pixel values.

We utilized a series of image-processing steps to analyse the distribution of pollen grains in different regions of the bee’s body. Initially, the original images were subjected to binarization (Supplementary Data Fig. S2) using GIMP (GNU Image Manipulation Program). In this process, pixels with one value represented pollen-occupied areas, while pixels with a zero value represented non-pollen-occupied areas. Following binarization, each image was manually aligned on Cartesian axes to accurately represent the bee’s size. Alignment was achieved using two tying points, i.e. reference location, at each lateral extremity of the bee’s body. This procedure allowed us to overlap all images in a 2-D space. To ensure accurate overlapping, we randomly selected one image per treatment as a model for cropping the other images. This step was accomplished using the image transformation tool for polygons, whereby the bee body was delimited in each treatment. The smoothed polygon was obtained through edge smoothing utilizing the polynomial approximation algorithm with an exponential kernel and a smoothing tolerance of 1 mm. The remaining images were subsequently cropped based on the smoothed polygon.

Next, we transformed each pixel occupied by pollen into a point and extracted the corresponding spatial x and y coordinates of each point. We used this information to produce heat maps of pollen placement for each treatment (Supplementary Data Fig. S3). We also estimated pollen density on the bee’s body by dividing the number of occupied pixels in each image by the total number of pixels available (i.e. the number of occupied pixels plus the number of unoccupied pixels). The image processing tasks for resampling the images, extracting pollen coordinates and estimating pollen density were performed using the Resample, Geoprocessing, Raster to Polygon, Raster to Point, Smooth Polygon, and Add XY Coordinates tools within ArcGIS Pro version 3.0.1.

Estimation of the inaccuracy components in the bee body

In our study, we estimated two components of the realized inaccuracy to assess the impact of the cucculus and stamen dimorphism on the position of pollen grains on the bee body relative to the position where the stigma touches. Inaccuracy is defined as the population deviation from a given optimum, being described by (1) the maladaptive bias, i.e. the mean departure from the optimum, and (2) the imprecision, i.e. the intrapopulation variation (sensu). To estimate the mean departure from the optimum, we utilized the coordinates of each pollen-occupied pixel in the digitized images of the bee body. Subsequently, we determined the distance between the average position of the pollen and the stigma optimum (i.e. the stigmas’ average coordinate where they touch the bee body). It is important to note that, as C. latistipula has enantiostyly with two floral morphs, the population has two stigma optima on the bee’s body. Therefore, the departure from the optimum was calculated considering the corresponding bee lateral region in each case. The distance measurement represents the first component of the inaccuracy, i.e. the maladaptive bias of pollen following each bee visit. In addition, we computed the sum of the standard deviations of the x and y coordinates of the pollen positions to estimate the second component of the inaccuracy, the realized intrapopulation variation in pollen placement. We chose to use the standard deviation rather than the variance proposed by due to our dataset’s absence of negative values and to achieve better model fit.

Statistical analysis

The data analysis was structured following the three key questions stated in the last paragraph of the Introduction.

Firstly, to investigate how the cucculus impacts the pollen deposition density in different regions of the bee’s body, we built a binomial generalized linear mixed model in which the response variable was the total number of pixels occupied by pollen on the bee’s body combined with the number of unoccupied pixels. The explanatory variables were categorical and included the condition of the cucculus (deflected or unmanipulated), the region of the bee’s body (lateral adjacent to the cucculus, dorsal, lateral opposite the cucculus, or ventral), as well as their interaction. For this analysis, we used data from treatments 1 and 4 (i.e. control flowers and flowers with deflected cucculus and both types of anthers unobstructed, respectively), totalling 252 analysed images. Plant identity was considered a random factor to control for morphological variation due to genetic basis, and a binomial error distribution with the logit link function was employed. After fitting the model, a type III ANOVA was conducted to determine the significance of each explanatory variable once we expected an interaction effect between the explanatory variables. Additionally, we compared the marginal means across different levels of the explanatory variable.

Secondly, to investigate the influence of different stamens on the pollen density placed on the lateral region of the bee’s body adjacent to the cucculus, we employed a binomial generalized linear mixed model, in which the response variable was also the total number of pixels occupied by pollen on the bee’s body, combined with the number of unoccupied pixels. The explanatory variable considered in this analysis was categorical in three levels: both types of anthers unobstructed (control), small or large anthers, both unobstructed. Here, we used data from treatments 1, 2 and 3 (i.e. all treatments involving unmanipulated cucculus, varying only the anthers condition), totalling 154 analysed images. Plant identity was also included as a random factor, and a binomial error distribution with the logit link function was applied. Following model adjustment, we conducted a type II variance analysis to assess each explanatory variable’s significance. Furthermore, we compared the marginal means across different levels of the explanatory variable. In this second analysis, we did not include data on bee vibrations in flowers with deflected cucculus. We conducted these two analyses separately rather than running a full model accounting for all six treatments because we previously knew that the other treatments produced many low or even zero pollen densities (Supplementary Data Fig. S3).

Thirdly, to investigate the roles of the cucculus and stamen dimorphism in the realized inaccuracy of pollen placement on the lateral region of the bee’s body, we estimated the two components of inaccuracy and used them as response variables in two different generalized mixed models. The explanatory variables used in the models were the condition of the cucculus (deflected or unmanipulated), the condition of the anthers (both types of anthers unobstructed, small anthers unobstructed, or large anthers unobstructed), and their interaction. Here, we considered all treatments but only the lateral side adjacent to the cucullus because this is the region where the stigma touches the bee’s body and showed the highest density of pollen grains. We considered a total of 182 images for each analysis. Plant identity was included as a random factor, and a Gaussian error distribution was assumed. We conducted a type III variance analysis to evaluate each explanatory variable’s significance. Additionally, we estimated the marginal means for posterior comparisons.

The data analyses were conducted using various packages in the R version 4.1.2 environment (http://www.r-project.org/). The specific packages used include ggplot2 (Wickham, 2016) for data visualization, glmmTMB (Brooks et al., 2017) for fitting generalized linear mixed models, emmeans (Russell et al., 2023) for estimating marginal means, car (Fox and Weisberg, 2019) for additional analyses and the DHARMa package (Hartig, 2024) for checking residual dispersion around the fitted models. We responsibly used ChatGPT to enhance the manuscript’s clarity, coherence and overall readability.

RESULTS

Effect of the cucculus on pollen deposition density on the bee’s body

The cucculus affects the placement of pollen grains on the body of artificial bees (Supplementary Data Fig. S3; Supplementary Data Table S2; Supplementary Data Video S1). Bees vibrated on flowers with an unmanipulated cucculus showed a higher density of pollen grains placed on their bodies than bees vibrated in flowers with a deflected cucculus (Fig. 2; Supplementary Data Table S2). However, this relationship depends on which region of the bee’s body is considered (Fig. 3; Supplementary Data Table S2; χ2 = 13021.76; d.f. = 3; P < 0.01). On the side of the bee body adjacent to the cucculus, unmanipulated flowers placed almost 10 times more pollen grains than flowers with deflected cucculus (Fig. 3; Supplementary Data Table S3; ratio = 9.88; z = 254.88; P < 0.01). In addition, flowers with an unmanipulated cucculus also increased by 2.3 times the density of pollen grains placed on the ventral region of the body of artificial bees (Fig. 3; Supplementary Data Table S3; ratio = 2.31; z = 79.09; P < 0.01). Notably, little to no pollen reached the bee’s dorsal and opposite lateral regions in our simulations, regardless of cucculus manipulation (Supplementary Data Fig. S3).

Fig. 2.

Fig. 2.

Open in a new tab

Pollen deposition density on the bee’s body. Increasingly warm colours highlight higher pollen densities until red. (A) and (B) are images of the lateral bee’s side adjacent to the cucculus deflected and unmanipulated, respectively. (C) and (D) are images of the ventral region of the bee’s body in flowers with cucculus deflected and unmanipulated, respectively.

Fig. 3.

Fig. 3.

Open in a new tab

Pollen deposition density on different regions of the artificial bee’s body. Pollen density was higher when the cucculus was left unmanipulated in both adjacent lateral and ventral regions. The artificial bee was built in a 3-D printer and coupled to a vibration transducer to simulate bee visits. Pollen density estimation was done using digital imaging techniques. *** indicates a significant contrast between estimated marginal means at a 95 % confidence interval.

Effect of stamen dimorphism on pollen density on the bee lateral side adjacent to the cucculus

The pollen density on the bee side adjacent to the unmanipulated cucculus depended on stamen dimorphism (Fig. 4; Supplementary Data Table S4; χ2 = 50055.00; d.f. = 2; P < 0.01). Flowers with all unobstructed anthers placed up to almost 3 times more pollen grains than other anther treatments (Fig. 4; Supplementary Data Table S5; ratioboth-only large = 4.55; z = 194.18; P < 0.01; ratioboth-only small = 2.85; z = 155.70; P < 0.01). Moreover, the pollen density deposited by small anthers was almost twice that of large anthers (Fig. 4; Supplementary Data Table S6; ratioonly small—only large = 1.59; z = 52.09; P < 0.01).

Fig. 4.

Fig. 4.

Open in a new tab

Pollen density deposited on the bee body side adjacent to the cucculus in flowers with unmanipulated cucculus under three conditions of anther pore manipulation. Both anthers were left unobstructed or the anthers of either large or small stamens were obstructed using superglue. Pollen density was higher when both anthers were unobstructed. The artificial bee used to vibrate the flowers was built in a 3-D printer and coupled to a vibration transducer to simulate bee visits. *** indicates a significant contrast between estimated marginal means at a 95 % confidence interval.

Effect of the cucculus and stamen dimorphism on the inaccuracy components of pollen grains

The two inaccuracy components were affected differently by the cucculus and stamen dimorphism. Firstly, the mean departure from the optimum was not modified by either the cucculus manipulation or the obstruction of anther types (Fig. 5A; Supplementary Data Table S6; χ2 = 1.29; d.f. = 2; P > 0.05). Therefore, the pollen positioning relative to the stigma does not depend on the cucculus or specifically on any anther type, although both strongly influence the pollen grain density. On the other hand, the intrapopulation variation (standard deviation) was impacted by manipulating the cucculus but not by anther obstruction. In this case, unmanipulated flowers placed more clumped pollen loads than flowers with deflected cucculus (Fig. 5B; Supplementary Data Table S7; χ2 = 239.61; d.f. = 1; P < 0.05), i.e. the deflection of the cucculus increased pollen placement imprecision by ~10 % on the bee’s body.

Fig. 5.

Fig. 5.

Open in a new tab

Components of the realized inaccuracy of pollen grains on the lateral region of the bee body adjacent to the cucculus. (A) Departure from the optimum (stigma position) across the cucculus and anther obstruction conditions. (B) Intrapopulation variation (standard deviation) across the cucculus and anther obstruction conditions. Flowers had the cucculus unmanipulated or manipulated and the two types of anthers unobstructed, only the small anthers obstructed or only the large anthers obstructed. Both departure from the optimum and intrapopulation variance were estimated from the coordinates of each occupied pixel in the digitized images. *** indicates a significant contrast between estimated marginal means at a 95 % confidence interval.

DISCUSSION

In this study we tracked the position of pollen grains on an artificial bee body, which allowed us to investigate the role of stamen dimorphism and the specialized petal called the cucculus of C. latistipula flowers in pollen placement, an essential step in the pollen’s journey towards the stigmas. We hypothesized that the cucculus and large stamens increase pollen density and placement accuracy on the bees’ bodies. In fact, our findings revealed that the cucculus enhances pollen deposition on the bee body region adjacent to it and on the bee’s venter. However, contrary to our expectations, small stamens contribute more pollen grains than large ones, always mediated by the cucculus, indicating a deviation from the typical function of stamen dimorphism in flowers with poricidal anthers. The contribution of smaller stamens to the pollination process explains the similar pollen viability across stamen types (Nogueira et al., 2018), in contrast to what has been previously reported in other studies (e.g. Saab et al., 2021). Additionally, the cucculus also influences pollen placement accuracy, reducing pollen placement imprecision on the bee’s body and guiding the pollen towards more favourable regions for reaching the stigmas of C. latistipula. Although ‘ricochet pollination’ has been documented in other flowers with a cucculus (Westerkamp, 2004; Amorim et al., 2017), this study provides the first quantitative estimation of its impact on pollen placement by both stamen types, including the smaller ones. This process is facilitated by modified petals, which, in turn, enhance the flower’s male function.

During buzz pollination in pollen flowers, both pollinator and plant can derive benefits based on the specific regions of the bee’s body where pollen is placed. Generally, pollen grains placed in easily accessible body regions can be collected through grooming behaviour, ensuring resources for the bee’s offspring (Thorp, 1979, 2000). On the other hand, pollen grains deposited on safe sites, which are located outside of grooming regions and in closer proximity to where the stigmas make contact, have a higher likelihood of effectively accomplishing pollination and subsequent fertilization of the ovules (Buchmann, 1983; Thorp, 2000; Koch et al., 2017; Tong and Huang, 2017). In such buzz-pollinated flowers, the division of the pollen load into different portions is primarily facilitated by different stamens within the flower – the stamen dimorphism that evolved in numerous lineages of pollen flowers (Vallejo-Marín et al., 2010; Melo et al., 2021). However, our findings demonstrate that the unequal distribution of pollen grains in these safe and unsafe sites on the bee’s body is primarily mediated by the modified petal cucculus in C. latistipula. By enhancing pollen deposition on the bee’s lateral body region adjacent to the cucullus, this petal increases pollen density on a safe site – where bees have limited access, but where stigmas of enantiostylous flowers frequently make contact during visits (Koch et al., 2017; Tong and Huang, 2017). Thus, by facilitating such a distribution of the pollen load, the cucculus primarily serves a male function during the pollination process. Besides its reproductive role, the cucculus also serves a feeding function, as evidenced by the higher density of pollen in the ventral region of the bee’s body when the cucculus was unaltered. These results highlight that the division of labour achieved solely through stamen dimorphism in most pollen flowers can be reinforced by specialized petals.

The evolution of offering only pollen as a floral reward for pollinators has been correlated with the existence of poricidal anthers and stamen dimorphism throughout the evolutionary history of angiosperms (Vallejo-Marín et al., 2010; Russell et al., 2024). In most flowers exhibiting these traits, larger stamens typically carry higher viable pollen loads, and their pollen grains are expected to have a greater chance of reaching the stigmas during pollination (Maia et al., 2019; Velloso et al., 2018; Dellinger et al., 2019; Brito et al., 2021; Saab et al., 2021; Trevisan et al., 2023). However, recent studies have challenged this well-established function of different stamens (Kay et al., 2020; Konzmann et al., 2020; Telles et al., 2020; Dellinger et al., 2021; Trevisan et al., 2023). For instance, in Clarkia unguiculata (Onagraceae), small stamens produce pollen that outperforms large stamens in both stigma and style penetration (Peach and Mazer, 2019). A similar pattern is observed in Pterolepis glomerata (Melastomataceae), in which small stamens produce more viable pollen and release more pollen grains during a single bee visit than large stamens (Telles et al., 2020). Interestingly, in the case of C. latistipula, the role of the cucculus contradicts our initial expectations, as the small stamens placed more pollen grains than the large stamens on the lateral region of the bee’s body. Thus, specialized petals in buzz-pollinated flowers can redirect pollen grains produced by the smaller stamens, which would primarily serve as a food source for bees, towards pollination. In fact, the anthers of the seven small stamens in this species seem to produce more pollen grains, ~64 % of the total pollen load per flower (A. Nogueira, unpubl. data), suggesting that the difference observed in pollen deposition may be just proportional to the total pollen produced by each stamen type. The proportional contribution of stamen types for pollination reinforces that the functional role of stamen types may be labile or even absent in some flowers with stamen dimorphism (Konzmann et al., 2020).

Pollen flowers, including C. latistipula, often exhibit enantiostyly, a floral system where the pistil is deflected to one side of the floral symmetry axis (Fidalgo et al., 2018). This deflection is commonly accompanied by the reciprocal positioning of stamens, corresponding to the stigma position of other flowers in the population, reducing the likelihood of self-pollination and promoting cross-pollination (Webb and Lloyd, 1986; Braga et al., 2022). Like heterostylous systems, enantiostyly can be considered a form of reciprocal herkogamy (Cardoso et al., 2018). In previous studies of heterostylous flowers, measures of inaccuracy have been used to assess the reciprocity between stamens and stigmas (Matias et al., 2020; Raupp et al., 2020; Trevizan et al., 2021; Braga et al., 2022). However, in our study we introduced a novel approach using both maladaptive bias and imprecision to estimate the impact of the cucculus and stamen dimorphism on the position of pollen grains relative to the stigmas on the bee’s body. Our findings demonstrate that the cucculus reduces the realized inaccuracy of pollen grains by decreasing their imprecision around the optimal position for stigmatic contact.

In those cases, specialized petals like the cucculus can enhance the deposition of pollen grains in optimal locations for effective pollination in enantiostylous flowers. However, it is important to interpret the effect of cucculus manipulation on pollen placement imprecision with caution, as the observed impact was relatively low. Additionally, it should be noted that in natural settings pollen grains deposited on the bodies of real pollinators, especially bees, may form 3-D layers that vary in their exposure to stigmatic contact (Price and Waser, 1982; Lertzman and Gass, 1983; Marcelo et al., 2022; Moir and Anderson, 2023). Since our study relied on 2-D photographs, our analysis did not consider the third dimension of pollen placement. The cucculus may decrease the realized inaccuracy of pollen grains by layering them on the bodies of actual bees. However, this hypothesis remains untested and would require further investigation.

Conclusions

Our study on C. latistipula revealed that the cucculus causes pollen grains to ricochet and be placed on the side adjacent to it and in the ventral region of the bee’s body. For the first time, we quantitatively estimated the redistribution and increased placement of pollen grains on the bee’s body, resulting from the presence of the cucculus. Furthermore, our findings indicate that pollen grains from both anther types, although especially from small anthers, can be directed towards plant reproductive purposes through their interaction with these specialized petals. By placing pollen grains on safe sites of voracious pollen collectors, commonly modified petals, such as the ones of the Cassieae tribe, plays a significant role in increasing the efficiency of pollen placement and, ultimately, pollen transfer. Our results highlighted the functional mechanism by which the modified petals enhance pollen placement on safe sites for plant sexual reproduction. In these cases, such specialized petals may have evolved mainly by increasing flowers’ male reproductive performance.

SUPPLEMENTARY DATA

Supplementary data are available at Annals of Botany online and consist of the following. Table S1: ANOVA type III table to compare the effect of cocculus manipulation on pollen placement by C. latistipula in different bee body regions. Table S2: estimated marginal means and contrast between treatments (manipulated or unmanipulated cucculus) based on the adjusted model created to compare the effect of cucculus manipulation on pollen placement by C. latistipula in different bee body regions. Table S3: estimated marginal means and contrasts between different bee body regions (lateral adjacent to the cucculus, dorsal, lateral opposite the cucculus, or ventral) based on the adjusted model created to compare the effect of cucculus manipulation on pollen placement by C. latistipula in different bee body regions. Table S4: ANOVA type II table to assess the effect of stamen dimorphism on pollen placement by C. latistipula flowers on the bee lateral side adjacent to the cucculus. Table S5: estimated marginal means and contrasts between different anther treatments (both anthers unobstructed, only large anthers unobstructed, or only small anthers unobstructed) based on the adjusted model created to compare the effect of stamen dimorphism on pollen placement by C. latistipula in the lateral region of the bee body adjacent to the cucculus. Table S6: ANOVA type III table to compare the effect of cucculus manipulation and stamen dimorphism on the maladaptive bias of pollen placement by C. latistipula in the lateral region of the bee’s body adjacent to the cucculus. Table S7: ANOVA type III table to compare the effect of cucculus manipulation and stamen dimorphism on the imprecision of pollen placement by C. latistipula in the lateral region of the bee’s body adjacent to the cucculus. Figure S1: experimental setup. Figure S2: following the vibration of each flower, the artificial bee was decoupled from the vibration transducer and was photographed from four different angles, emphasizing its lateral region adjacent to the cucculus (A), dorsal region (C), lateral region opposite to the cucculus (E) and ventral region (G), while placed on a black foam stage. Figure S3: density of pollen placed on the four regions of the bee’s body. Video S1: video showing the effect of the cucculus on pollen deposition density on the bee’s body.

mcae210_suppl_Supplementary_Figures_S1-S3_Tables_S1-S8
mcae210_suppl_supplementary_figures_s1-s3_tables_s1-s8.docx (248.9KB, docx)
mcae210_suppl_Supplementary_Video_S1
Download video file (44.9MB, wmv)

ACKNOWLEDGEMENTS

We thank Raphael Matias da Silva, Nathália Susin Streher and two anonymous reviewers for their valuable comments on the previous versions of the manuscript. We also thank everyone who assisted us in the experiment and data collection during the various stages of this study. In particular, we thank Amanda Vieira da Silva and Bruna Campos Barbosa for their help during the experiment and data collection. Additionally, we thank the Multiuser Center for Biodiversity and Conservation (CMBC‐PROPES) at UFABC for generously providing the facilities for plant cultivation and executing experimental procedures.

Contributor Information

Thainã R Monteiro, Programa de pós-graduação em Ecologia, Conservação e Biodiversidade, Instituto de Biologia, Universidade Federal de Uberlândia, Uberlândia, 38405-315, Brazil.

Rogério V S Gonçalves, School of Earth, Atmospheric and Life Sciences, University of Wollongong, Wollongong, 2522, Australia.

Francismeire J Telles, Programa de pós-graduação em Ecologia, Conservação e Biodiversidade, Instituto de Biologia, Universidade Federal de Uberlândia, Uberlândia, 38405-315, Brazil.

Gudryan J Barônio, Departamento de Ecologia, Instituto de Biociências, Universidade de São Paulo (IB-USP), São Paulo, SP, 05508-090, Brazil.

Anselmo Nogueira, Laboratório de Interações Planta-Animal, Centro de Ciências Naturais e Humanas, Universidade Federal do ABC, São Bernardo do Campo, 09606-045, Brazil.

Vinícius L G Brito, Instituto de Biologia, Universidade Federal de Uberlândia, Uberlândia, 38405-315, Brazil.

FUNDING

This work was supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES 001, grants to T.R.M., R.V.S.G. and F.J.T.); Fundação de Amparo à Pesquisa do Estado de São Paulo (grants 2021/09247-5 to G.J.B. and 2019/19544-7 to A.N.); Fundação de Amparo à Pesquisa do Estado de Minas Gerais (grants RED-00253-16, RED-00039-23 and APQ-01586-24) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (grants 308107/2021-7, 401843/2022-0 and 350145/2022-9 to V.L.G.B., 312389/2023‐0 to A.N. and 423939/2021-1 to all).

DATA AVAILABILITY

All the data used in this research will be archived correctly in the Dryad repository (https://datadryad.org/stash) upon acceptance for publication.

LITERATURE CITED

  1. Almeida  NM, Bezerra  TT, Oliveira  CRS, et al.  2015a. Breeding systems of enantiostylous Cassiinae species (Fabaceae, Caesalpinioideae). Flora  215: 9–15. [Google Scholar]
  2. Almeida  NM, Cotarelli  VM, Souza  DP, et al.  2015b. Enantiostylous types of Cassiinae species (Fabaceae-Caesalpinioideae). Plant Biology (Stuttgart)  17: 740–745. [DOI] [PubMed] [Google Scholar]
  3. Amorim  T, Marazzi  B, Soares  AA, Forni-Martins  ER, Muniz  CR, Westerkamp  C.  2017. Ricochet pollination in Senna (Fabaceae) – petals deflect pollen jets and promote division of labour among flower structures. Plant Biology  19: 951–962. [DOI] [PubMed] [Google Scholar]
  4. Arroyo-Correa  B, Beattie, C, Vallejo-Marín  M. 2019. Bee and floral traits affect the characteristics of the vibrations experienced by flowers during buzz pollination. Journal of Experimental Biology  222. [DOI] [PubMed] [Google Scholar]
  5. Braga  BLP, Matias  R, Consolaro  H, Souza  JT, Almeida  NM.  2022. Inaccuracy of sexual organ position and spatial variation of styles in monomorphic enantiostylous flowers of Senna rugosa. Flora  293: 152112. [Google Scholar]
  6. Brito  VLG, Nunes  CEP, Resende  CR, Montealegre-Zapata  F, Vallejo-Marín  M.  2020. Biomechanical properties of a buzz-pollinated flower. Royal Society Open Science  7: 201010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Brito  VLG, Leite  FB, Jorge  LR, Sazima  M.  2021. Distinct pollen release dynamics between stamens generate division of labour in pollen flowers of two Pleroma species (Melastomataceae). Flora  285: 151961. [Google Scholar]
  8. Brooks  ME, Kristensen  K, Benthem  KJV, et al.  2017. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R Journal  9: 378–400. [Google Scholar]
  9. Buchmann  SL.  1983. Buzz pollination in angiosperms. In: Jones  CE, Little  RJ. eds. Handbook of experimental pollination biology.  New York: Van Nostrand Reinhold, 73–113. [Google Scholar]
  10. Camargo  RA, Miotto  STS.  2004. O gênero Chamaecrista Moench (Leguminosae-Caesalpinioideae) no Rio Grande do Sul. Iheringia Serie Botanica  59: 131–148. [Google Scholar]
  11. Cardoso  JCF, Viana  ML, Matias  R, et al.  2018. Towards a unified terminology for angiosperm reproductive system. Acta Botanica Brasilica  32: 329–348. [Google Scholar]
  12. Delgado-Salinas  AO, Sousa-Sánchez  M.  1977. Floral biology of the genus Cassia in the region of Los Tuxtlas, Veracruz. Botanical Sciences  37: 5–52. [Google Scholar]
  13. Dellinger  AS, Pöllabauer  L, Loreti  M, Czurda  J, Schönenberger  J.  2019. Testing functional hypotheses on poricidal anther dehiscence and heteranthery in buzz-pollinated flowers. Acta ZooBot Austria  156: 197–214. [Google Scholar]
  14. Dellinger  AS, Artuso  S, Fernández-Fernández  DM, Schönenberger  J.  2021. Stamen dimorphism in bird-pollinated flowers: investigating alternative hypotheses on the evolution of heteranthery. Evolution  75: 2589–2599. [DOI] [PubMed] [Google Scholar]
  15. de Luca  PA, Bussière  LF, Souto-Vilaros  D, Goulson  D, Mason  AC, Vallejo-Marín  M.  2013. Variability in bumblebee pollination buzzes affects the quantity of pollen released from flowers. Oecologia  172: 805–816. [DOI] [PubMed] [Google Scholar]
  16. Fidalgo  A de O, Guimarães  DM, Caldiron  GT, Barbosa  JM.  2018. Reproductive ecology of two pioneer legumes in a coastal plain degraded by sand mining. Hoehnea  45: 93–102. [Google Scholar]
  17. Fox  J, Weisberg  S.  2019. An R companion to applied regression, 3rd edn. Thousand Oaks: Sage. [Google Scholar]
  18. Hartig  F.  2024. DHARMa: residual diagnostics for hierarchical (multi-level/mixed) regression models. Theoretical Ecology, University of Regensburg, R package version 0.4.7, https://github.com/florianhartig/dharma. [Google Scholar]
  19. Kay  KM, Jogesh  T, Tataru  D, Akiba  S.  2020. Darwin’s vexing contrivance: a new hypothesis for why some flowers have two kinds of anther: heteranthery in Clarkia. Proceedings of the Royal Society B: Biological Sciences  287: 20202593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Koch  L, Lunau  K, Wester  P.  2017. To be on the safe site – ungroomed spots on the bee’s body and their importance for pollination. PLoS One  12: e0182522. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Konzmann  S, Hilgendorf  F, Niester  C, Rech  AR, Lunau  K.  2020. Morphological specialization of heterantherous Rhynchanthera grandiflora (Melastomataceae) accommodates pollinator diversity. Plant Biology (Stuttgart)  22: 583–590. [DOI] [PubMed] [Google Scholar]
  22. Lertzman  KP, Gass  CL.  1983. Alternative models of pollen transfer. In: Jones  CE, Little  RJ. eds. Handbook of Experimental Pollination Biology. New York: Van Nostrand Reinhold, 474–489. [Google Scholar]
  23. Li  JK, Song  YP, Xu  H, Zhang  YW, Zhu  JY, Tang  LL.  2015. High ratio of illegitimate visitation by small bees severely weakens the potential function of heteranthery. Journal of Plant Ecology  8: 213–223. [Google Scholar]
  24. Lunau  K.  2004. Adaptive radiation and coevolution – pollination biology case studies. Organisms Diversity and Evolution  4: 207–224. [Google Scholar]
  25. Luo  ZL, Gu  L, Zhang  DX.  2009. Intrafloral differentiation of stamens in heterantherous flowers. Journal of Systematics and Evolution  47: 43–56. [Google Scholar]
  26. Luo  Z, Zhang  D, Renner  SS.  2008. Why two kinds of stamens in buzz-pollinated flowers? Experimental support for Darwin’s division-of-labour hypothesis. Functional Ecology  22: 794–800. [Google Scholar]
  27. Maia  KP, Vaughan  IP, Memmott  J.  2019. Plant species roles in pollination networks: an experimental approach. Oikos  128: 1446–1457. [Google Scholar]
  28. Marazzi  B, Endress  PK.  2008. Patterns and development of floral asymmetry in Senna (Leguminosae, Cassiinae). American Journal of Botany  95: 22–40. [DOI] [PubMed] [Google Scholar]
  29. Marcelo  V, Marquitti  FMD, Vallejo-Marín  M, Brito  VLG.  2022. Pollinator grooming behavior alters pollen landscapes on bees’ bodies and increases pollen carryover to other flowers. bioRxiv. [Preprint] doi: https://doi.org/ 10.1101/2022.07.14.499901. [DOI]
  30. Matias  R, Pérez-Barrales  R, Consolaro  H.  2020. Patterns of variation in distylous traits and reproductive consequences in Erythroxylum species and populations. American Journal of Botany  107: 1–13. [DOI] [PubMed] [Google Scholar]
  31. Melo  LRF, Vasconcelos  T, Reginato  M, Caetano  APS, Brito  VLG.  2021. Evolution of stamen dimetrism in Melastomataceae, a large radiation of pollen flowers. Perspectives in Plant Ecology, Evolution and Systematics  48: 125589. [Google Scholar]
  32. Mesquita-Neto  JN, Costa  BKP, Schlindwein  C.  2017. Heteranthery as a solution to the demand for pollen as food and for pollination – Legitimate flower visitors reject flowers without feeding anthers. Plant Biology  19: 942–950. [DOI] [PubMed] [Google Scholar]
  33. Moir  M, Anderson  B.  2023. Pollen layering and male–male competition: quantum dots demonstrate that pollen grains compete for space on pollinators. American Journal of Botany  2022: 1–9. [DOI] [PubMed] [Google Scholar]
  34. Müller  H.  1881. Two kinds of stamens with different functions in the same flower. Nature  24: 307–308. [Google Scholar]
  35. Müller  H.  1882. Two Kinds of Stamens with Different Functions in the same Flower. Nature  27: 30. [Google Scholar]
  36. Müller  H.  1883. The fertilization of flowers. (transl. by Thompson DW). London: Macmillan and Co. [Google Scholar]
  37. Nogueira  A, Valadão-Mendes  LB, El Ottra  JHL, et al.  2018. Relationship of floral morphology and development with the pattern of bee visitation in a species with pollen-flowers, Chamaecrista desvauxii (Fabaceae). Botanical Journal of the Linnean Society  187: 137–156. [Google Scholar]
  38. Paulino  JV, Mansano  VF, Prenner  G.  2016. Evidence for division of labor and division of function related to the pollen release in Papilionoideae (Leguminosae) with a heteromorphic androecium. International Journal of Plant Sciences  177: 590–607. [Google Scholar]
  39. Paulino  JV, Mansano  VF, Teixeira  SP.  2013. Elucidating the unusual floral features of Swartzia dipetala (Fabaceae). Botanical Journal of the Linnean Society  173: 303–320. [Google Scholar]
  40. Peach  K, Mazer  SJ.  2019. Heteranthery in Clarkia: pollen performance of dimorphic anthers contradicts expectations. American Journal of Botany  106: 598–603. [DOI] [PubMed] [Google Scholar]
  41. Price  MV, Waser  NM.  1982. Experimental studies of pollen carryover: hummingbirds and Ipomopsis aggregata. Oecologia  54: 353–358. [DOI] [PubMed] [Google Scholar]
  42. Pritchard  DJ, Vallejo-Marín  M.  2020. Floral vibrations by buzz-pollinating bees achieve higher frequency, velocity and acceleration than flight and defence vibrations. Journal of Experimental Biology  223: jeb220541. [DOI] [PubMed] [Google Scholar]
  43. Rando  JG, Cota  MMT, Conceição  AS, Barbosa  AR, Barros  TLA.  2021. Chamaecrista in Flora do Brasil 2020.  Jardim Botânico do Rio de Janeiro. [Google Scholar]
  44. Rando  JG, Cota  MMT, Lima  AG, Bortolozzi  RLC, Marazzi  B, Conceição  AS.  2024. 5. Tribe Cassieae. In: Bruneau A, Queiroz LP, Ringelberg JJ, eds. Advances in Legume Systematics 14. Classification of Caesalpinioideae. Part 2: higher-level classification. PhytoKeys  240: 83–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Raupp  PP, Matias  R, Furtado  MT, Consolaro  H.  2020. The role of distyly in pollen flow of the hummingbird-pollinated Palicourea rigida (Rubiaceae). Flora  271: 151681. [Google Scholar]
  46. Renner  SS.  1990. A revision of Rhynchanthera (Melastomataceae). Nordic Journal of Botany  9: 601–630. [Google Scholar]
  47. Ruhren  S.  2003. Seed predators are undeterred by nectar-feeding ants on Chamaecrista nictitans (Caesalpineaceae). Plant Ecology  166: 189–198. [Google Scholar]
  48. Russel  AL, Zenil-Ferguson  R, Buchmann  SL, Jolles  DD, Kriebel  R, Vallejo-Marín  M.  2024. Widespread evolution of poricidal flowers: a striking example of morphological  convergence across flowering plants. bioRxiv [Preprint]. doi: https://doi.org/ 10.1101/2024.02.28.582636. [DOI] [Google Scholar]
  49. Russell  A, Lenth  V, Bolker  B, et al.  2023. Package ‘emmeans’ R topics documented. 34: 216–221. [Google Scholar]
  50. Saab  GDS, Mansano  VDF, Nogueira  A, Maia  IC, Bergamo  PJ, Paulino  JV.  2021. A sophisticated case of division of labour in the trimorphic stamens of the Cassia fistula (Leguminosae) flower. AoB Plants  13: 1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Telles  FJ, Klunk  CL, Maia  FR, Brito  VLG, Varassin  IG.  2020. Towards a new understanding of the division of labour in heterantherous flowers: the case of Pterolepis glomerata (Melastomataceae). Biological Journal of the Linnean Society  131: 1–11. [Google Scholar]
  52. Thorp  RW.  1979. Structural, behavioral and physiological adaptations of bees (Apoidea) for collecting pollen. Annals of the Missouri Botanical Garden  66: 788–812. [Google Scholar]
  53. Thorp  RW.  2000. The collection of pollen by bees. Plant Systematics and Evolution  222: 211–223. [Google Scholar]
  54. Tong  ZY, Huang  SQ.  2017. Safe sites of pollen placement: a conflict of interest between plants and bees? Oecologia  186: 163–171. [DOI] [PubMed] [Google Scholar]
  55. Trevizan  R, Cardoso  JCF, Coelho  CP, Consolaro  H, Furtado  MT, Oliveira  PE.  2021. Inaccuracy patterns and precise pollination among distylous Palicourea and Psychotria (Rubiaceae). Perspectives in Plant Ecology, Evolution and Systematics  53: 125645. [Google Scholar]
  56. Trevizan  R, Caetano  APS, Brito  VLG, Oliveira  PE, Telles  FJ.  2023. Stamen and pollen heteromorphism linked to the division of labour in Melastomataceae species. Flora  305: 152315. [Google Scholar]
  57. Vallejo-Marín  M.  2019. Buzz pollination: studying bee vibrations on flowers. The New Phytologist  224: 1068–1074. [DOI] [PubMed] [Google Scholar]
  58. Vallejo-Marín  M, Manson  JS, Thomson  JD, Barrett  SCH.  2009. Division of labour within flowers: heteranthery, a floral strategy to reconcile contrasting pollen fates. Journal of Evolutionary Biology  22: 828–839. [DOI] [PubMed] [Google Scholar]
  59. Vallejo-Marín  M, Da Silva  EM, Sargent  RD, Barrett  SCH.  2010. Trait correlates and functional significance of heteranthery in flowering plants. The New Phytologist  188: 418–425. [DOI] [PubMed] [Google Scholar]
  60. Vallejo-Marín  M, Russell  AL.  2024. Harvesting pollen with vibrations: towards an integrative understanding of the proximate and ultimate reasons for buzz pollination. Annals of Botany  133: 379–398. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Velloso  MSC, Brito  VLG, Caetano  APS, Romero  R.  2018. Anther specializations related to the division of labor in Microlicia cordata (Spreng.) Cham. (Melastomataceae). Acta Botanica Brasilica  32: 349–358. [Google Scholar]
  62. Vogel  S.  1978. Evolutionary shifts from reward to deception in pollen flowers. In: Richards  AJ. ed. The pollination of flowers by insects.  London: Academic Press, 89–96. [Google Scholar]
  63. Webb  CJ, Lloyd  DG.  1986. The avoidance of interference between the presentation of pollen and stigmas in angiosperms II. Herkogamy. New Zealand Journal of Botany  24: 163–178. [Google Scholar]
  64. Westerkamp  C.  1996. Pollen in bee-flower relations. Some considerations on melittophily. Botanica Acta  109: 325–332 [Google Scholar]
  65. Westerkamp  C.  1997. Flowers and bees are competitors - Not partners. Towards a new understanding of complexity in specialised bee flowers. Acta Horticulturae. https://doi.org/ 10.17660/ActaHortic.1997.437.5 [DOI] [Google Scholar]
  66. Westerkamp  C.  2004. Ricochet pollination in cassias – and how bees explain enantiostyly. In: Freitas  MB, Pereira  JOP. eds. Conservation, rearing and management for pollination  Fortaleza: Imprensa Universitaria, 223–230. [Google Scholar]
  67. Wickham  H.  2016. ggplot2: elegant graphics for data analysis.  New York: Springer. https://ggplot2.tidyverse.org. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mcae210_suppl_Supplementary_Figures_S1-S3_Tables_S1-S8
mcae210_suppl_supplementary_figures_s1-s3_tables_s1-s8.docx (248.9KB, docx)
mcae210_suppl_Supplementary_Video_S1
Download video file (44.9MB, wmv)

Data Availability Statement

All the data used in this research will be archived correctly in the Dryad repository (https://datadryad.org/stash) upon acceptance for publication.

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES