Subtle ecophysiological divergences in a deceptive strategy to attract the same pollinators in two sympatric Arum

Abstract

Background and Aims

Integrative approaches are mandatory to comprehend the evolution of deceptive pollination strategies, by combining the co-variations in different floral traits and the behaviour of the insects. We applied an ecophysiological approach to compare floral traits (scent and thermogenesis) in two deceptive sympatric Arum that are pollinated mainly by Psychoda moth flies. We hypothesized that floral traits diverged between the two arums to minimize their interference in attraction of pollinators.

Methods

We measured several floral traits, including the timing of anthesis, thermogenesis patterns and floral volatile organic compounds, in Arum italicum and Arum maculatum. Additionally, we conducted laboratory experiments to determine the diel activity rhythm of the moth fly Psychoda sigma. We compared the activity window of the fly with the timing of anthesis, and we compared the two arums in the ability of their volatiles to increase pollinator activity.

Key Results

The main thermogenesis peak involved in pollinator attraction was cooler and occurred slightly earlier during the day in A. maculatum compared with A. italicum. The fly P. sigma appeared to be nocturnal, hence inflorescences emitting their floral scent before sunset are less likely to be visited by pollinators owing to the relatively short attractive (few hours) female phase. Regarding scents, we confirmed a strong differentiation between the two Arum species. Behavioural assays with P. sigma revealed a slightly stronger stimulation to the odour of A. maculatum compared with A. italicum.

Conclusions

Our integrative approach highlights that the two Arum species exhibit slightly shifted daily timing of anthesis, indicating a fine-tuned variation in their attractive volatiles and thermogenesis sequences in relationship to pollinator behaviour. A trade-off might exist between odour specificity and the ability to maximize volatile emission. These findings illustrate how evolutionarily close and sympatric species have shifted their attraction systems, potentially reducing competition within sympatric populations.

INTRODUCTION

Numerous plant taxa rely on deceptive methods to attract their pollinators by mimicking signals emitted by rewarding plants or by resources important to pollinators, such as food sources, mating and brood sites (Dafni, 1984; Internicola and Harder, 2012; Bohman et al., 2016). The Araceae family ranks second after Orchidaceae in the number of deceptive species (Chartier et al., 2014). Chemical mimicry of the brood site, using volatile organic compounds (VOCs), is one of the most widespread deceptive strategies in Araceae (Urru et al., 2011). Notably, species from various genera in the tribe Areae, including Arum, mimic the scents of carrion or decomposing organic material, most often attracting saprophilous or necrophilous insects (Angioy et al., 2004; Laina et al., 2022; Szenteczki et al., 2022).

In some taxa, deceptive mechanisms are enhanced further by a second mechanism: thermogenesis. Thermogenesis is a costly floral trait found in many plant species across both gymnosperms (Roemer et al., 2005) and angiosperms (Maekawa et al., 2022). The function and mechanisms remain unclear (Ervik and Barfod, 1999), but several hypotheses (non-mutually exclusive) were proposed, and most of them are linked to the interaction with pollinators (Van der Kooi et al., 2019). Clear demonstrations of these functions remain rather rare: the heat reward for tropical scarab beetles pollinating a Philodendron species (Seymour et al., 2003); pollen germination and growth stimulation in the temperate Symplocarpus renifolius (Seymour et al., 2009); or direct effect to lure carrion blowflies in complement to colour and scent in Helicodiceros muscivorus (Angioy et al., 2004). In temperate arums, the timing of heat production matches the timing of emission of VOCs from the flower (Marotz-Clausen et al., 2018), suggesting that thermogenesis might enhance the diffusion of these chemicals (Seymour, 2010), thereby increasing the range of attraction for pollinators.

Nevertheless, to what extent the pattern of heat production in thermogenic flowers (e.g. maximum temperature, duration or heating rate) co-varies with the VOCs (inter- or intraspecifically) remains an open question. Moreover, the complexity and regulation of VOCs and heat production patterns is useless if the system fails to attract pollinators. This pollination process requires synchronization between the biological rhythms of both the flower and its pollinators. Thus, the period of emission of VOCs needs to align with the daily timing when pollinators are active, i.e. the diel activity patterns (Guerenstein et al., 2004; Friberg et al., 2014). However, to our knowledge, the relationship between daily pollinator activity, flower temperature patterns and emission of VOCs remains underexplored in deceptive pollination systems.

Given the complexity of these attraction mechanisms, integrative approaches combined with fine-scale analyses of the processes involved are mandatory (Terry et al., 2007). The temperate arums Arum italicum and Arum maculatum offer promising biological systems for studying these mechanisms. Common across Europe, they grow in the understorey of deciduous forests. The main and hottest thermogenic organ (i.e. the appendix) warms once during the scent emission. The heat-production patterns (Fig. 1; Albre et al., 2003), production of VOCs (Kite, 1995; Gfrerer et al., 2021; Szenteczki et al., 2022) and pollinator communities (Gibernau et al., 2004; Chartier et al., 2016; Szenteczki et al., 2021; Gfrerer et al., 2023) for the two species have already been described and studied. However, these processes have mainly been studied separately, and few integrative studies have illustrated the fine-tuning between emission of VOCs, thermogenesis and the diel activity of pollinators (Marotz-Clausen et al., 2018).

Fig. 1.

Theoretical thermogenesis cycle of Arum italicum. The temperature was recorded during the 3 days of anthesis, including thermogenesis peaks in stamens (P1, P2 and P4) and in the appendix (P3) corresponding to the emission of volatile organic compounds. P3st is stamen temperature at the time of P3 (adapted from Albre et al., 2003).

The scents of A. italicum and A. maculatum are similar in their composition, mostly varying in the relative proportions of VOCs, although some compounds are not shared between them (Chartier et al., 2013; Gfrerer et al., 2021). This variation in VOCs explains the differences in species pollinator assemblages between the two Arum species, although Psychoda is the most representative genus in both (Chartier et al., 2016). Several VOCs also mimic the odour of horse and cow dung, which is the oviposition substrate of their main pollinator, Psychoda (Satchell, 1947; Frank et al., 2018; Gfrerer et al., 2022), suggesting the mimicry of floral VOCs to attract moth flies (Chartier et al., 2011, 2016). This deceptive pollination strategy also includes floral thermogenesis synchronized with anthesis, probably to enhance scent dispersion (Gibernau et al., 2004), although the function of certain thermogenic peaks remains unknown.

The short (few hours) visitor attraction period suggests co-evolution between flowering and pollinator activity, as seen in other flower–pollinator pairs (Hoballah et al., 2005; Jürgens et al., 2014; Balducci et al., 2020). Although A. italicum and A. maculatum are known to emit scents and attract pollinators in the evening (Lack and Diaz, 1991; Laina et al., 2022), no empirical evidence has yet been provided on the daily activity patterns of Psychoda (the main pollinators) or their synchronization with floral scent emission. Additionally, given that A. italicum is more generalist in its pollinator attraction, it remains unclear whether female moth flies (Psychoda) respond in a similar manner to the scents of both Arum species.

Our aim was to detail, for the first time, the relationship between the inflorescence characteristics (floral VOCs and thermogenesis) and pollinator behaviour in pollination ecology by matching: (1) the thermogenesis patterns of the two sympatric species (A. italicum and A. maculatum); (2) the floral scents emitted when the two plants co-occur and their correlation with flower temperatures; and (3) the diel activity of the main pollinator and its stimulation levels by the two respective flower scents. Strong competition for pollinators can be expected when the two species are sympatric, because each pollinator is trapped for 1 day when it is attracted by one inflorescence, making the pollinator unavailable to other plants. This comparative approach addresses whether the two Arum species have different attraction processes and whether the anthesis is synchronized with the activity time of their main pollinator. We hypothesized that inflorescences exhibit different scent patterns between the two species at our study site, with many shared VOCs but in different proportions, consistent with a previous study at a Pyrenean site (Chartier et al., 2016). We also expected slightly different interspecific thermogenesis patterns, because A. maculatum has smaller inflorescences (Chartier and Gibernau, 2009), implying different physical capacities. We also tested the hypothesis of synchronization between the daily cycle of pollinator activity and the attraction phase in the two Arum species. Finally, we hypothesized that the VOCs from the two Arum species stimulate the activity of Psychoda females differently, in relationship to their level of specificity towards each of the two Arum inflorescences (Chartier et al., 2016).

MATERIALS AND METHODS

Biological materials and field site

The two Arum species, A. italicum and A. maculatum, use thermogenesis coupled with scent emission to attract their pollinators. Pollinators are attracted and trapped in a floral chamber containing the reproductive flowers for 1 day, allowing the pollination of protogynous inflorescences. Here, individuals were sampled from a sympatric population at Montlouis-sur-Loire in France (47°22′53.5″N, 0°50′11.1″E) between 12 and 28 April 2021. The individuals were sampled within a 400 m radius in a semi-urban deciduous forest where the vegetation cover was similar for both species, thereby eliminating potential microhabitat effects. Time measurements were performed using the UTC+2 system.

The pollinators, moth flies of the Psychoda genus (Diptera), were collected from the inflorescences in April 2021, at this same study site. We found mostly two species of Psychoda in the inflorescences of Arum, namely Psychoda sigma and Psychoda uncinula (Leclerc et al., 2022). Species determination was carried out under a binocular stereomicroscope using the number of antennal segments and the shape of the genitalia (Quate, 1954; Withers, 1989; Ježek, 1990; Ježek and van Harten, 2005), and identifications were confirmed by an international specialist on the tribe, Gunnar M. Kvifte (Nord University, Norway). For behavioural studies, we used P. sigma, because breeding was more efficient and it represented a significant proportion of the pollinators observed in the inflorescences on our experimental site (M. A. J. Leclerc, pers. obs.). This Psychoda species was reared in the laboratory on horse dung in a climatic chamber at 20 °C following the same procedure as described by Leclerc et al. (2022).

Scent analysis

The floral scents (n = 5 for A. italicum and n = 6 for A. maculatum) were collected in the field in the evening of anthesis between 17:30 and 20:00 h, by solid phase micro-extraction [SPME fibres, 57328-U (DVB/CAR/PDMS); Supelco^®] following the method described by Chartier et al. (2016). The inflorescences were placed in an open headspace (colourless Nalophan NA; Rôti-bulle^®), closing the bottom with string. SPME fibre was attached and placed at 1–2 cm from the appendix for 45 min. As a control, a fibre was placed ∼4 m away from inflorescences to identify background odours. The fibres were stored at −20 °C until analyses by gas chromatography coupled to a mass spectrometer (7890B GC-7000C MS, Triple Quad, Agilent Technologies, Santa Clara, CA, USA).

The fibres were desorbed at 250 °C during 5 min into the injector in splitless mode. Helium was the carrier gas with a flow of 1.2 mL min⁻¹. The column (Zebron ZH-5HT inferno, 30 m × 0.25 mm × 0.25 μm, Phenomenex, Torrance, CA, USA) temperature program was as follows: the initial temperature was 50 °C with a ramp rate of 3 °C min⁻¹ to 200 °C, then the rate increased to 10 °C min⁻¹ until 225 °C and 5 min final hold at 225 °C. The identifications of the compounds was performed using NIST software (v.2.3, 2017) by comparing the Kováts’ retention indices (Kováts and Weisz, 1965) and the spectral profiles available in the libraries Adams (2011), Mainlib and Wiley8 (Petersen and Blom, 2011). Redundant compounds between the control and the flower samples were removed from the analyses. Peak integrations were performed using the MS Quantitative Analysis software (Agilent Technology, v.10.2) after characterization of all the peaks of interest (threshold of 1 % of the total relative amount; Supplementary Data Table S1; Fig. S1). The main compounds were characterized by a threshold of 5 % of mean relative proportion.

Thermogenesis

Floral temperatures were measured continuously (every 5 min) during the 3 days of anthesis (n = 11 for A. italicum and n = 12 for A. maculatum) including the plants used for scent measurements. We used the loggers HOBO UX120-014M (Onset Computer Corporation, USA) equipped with three type-T thermocouples (diameter 0.2 mm, copper/constantan, TC S.A., Dardilly, France) to record air and inflorescence temperatures simultaneously. Before the spathe opened, one thermocouple was inserted into the middle of the appendix and another into the stamens. The air temperature was measured at 0.5 m from the inflorescence at the same height.

Floral temperature patterns were analysed at several key moments (Fig. 1; for a detailed version, see Supplementary Data Fig. S2). Excess temperature was calculated to normalize flower temperatures according to air temperature (i.e. flower minus air temperature). After anthesis, the size of the appendix was measured and its volume calculated, assuming its shape to be a cylinder, to evaluate the heat production normalized by the appendix volume (in degrees Celsius per centimetre cubed) as a proxy for energetic ‘effort’. The duration and the maximum excess temperature from organs were also examined, and the detailed results are provided in the supplementary files for clarity and conciseness (Supplementary Data Results).

Pollinator activity

In the first experiment, we determined the diel activity pattern of P. sigma. The emerging individuals were sexed and transferred individually into closed tubes (Supplementary Data Fig. S3). The tubes were positioned in the grid of a Locomotor Activity Monitor system (LAM25H, TriKinetics Inc., Waltham, MA, USA) managed by the software DAMSystem3 Data Collection Software (v.3.08). The activity was measured by the cumulative counts of crossings of an infrared barrier in the middle of the tube (Liu et al., 2015) every 5 min during 24 h (all counts were summed per hour). The light system of the climatic chamber consisted of two LED bars of 4 W each (4000 K, 400 lm; BKL1159, BKLicht GmbH & Co. KG). Recordings began 15 min after the individuals were placed in a climatic chamber, at 20 °C, to allow time for acclimation. We tested 17 individuals of both sexes in normal light conditions (14 h light–10 h dark), 18 females and 23 males in constant dark conditions (24 h dark), and 15 individuals of both sexes in constant light conditions (24 h light). The constant light treatments were used to determine the presence of an endogenous biological clock.

In the second experiment, the impact of arum scents on the female P. sigma activity was measured during the first hour of darkness, corresponding to the time of the floral scent release in the field. Owing to the complexity floral odours, only the five main and most representative compounds of the two Arum species were selected, based on our chemical analyses (see Results below) to create simplified and reproductible floral scents in the laboratory. The concentration and quantity of the compounds was determined using our results and a previous study by Gfrerer et al. (2021). Pentane (5 µL, n = 21) was chosen as the solvent and negative control according to Stökl et al. (2010). For A. italicum (n = 24), 5 µL of solution was deposited on a filter paper strip (HVLP04700, 0.45 µm; Durapore^®) with the following VOC concentrations: 9.2 ng µL⁻¹ of 3,7-dimethyl-1-octene, 12 ng µL⁻¹ of β-citronellene and 3.6 ng µL⁻¹ of β-caryophyllene. For A. maculatum (n = 22), 5 µL of solution was deposited with a concentration of 14 ng µL⁻¹ of 2-heptanone, 30 ng µL⁻¹ of p-cresol, 28 ng µL⁻¹ of indole and 12 ng µL⁻¹ of α-humulene. The simplified floral odours or controls (pentane or 30 mg of fresh horse dung, n = 22) were placed in a micropipette cone and sent into the experimental tube. Three successive puffs of air with a 2 mL Pasteur pipette were used before darkness to detect any diurnal stimulation by odour during the 15 min before dark.

Statistical analyses

All statistical analyses were performed with the software R (v.4.0.3; R Development Core Team, 2020). Model requirements were verified using DHARMa package (v.0.4.6; Hartig and Lohse, 2022) and the functions simulateResiduals and testResiduals. The significance of the variables was tested with the Anova function of the Car package (v.3.0-10; Fox et al., 2022). The pairwise comparisons were performed with the emmeans function (package emmeans v.1.8.5; Lenth et al., 2023).

Scent analysis

Unconstrained redundancy analysis was applied to visualize the interspecific variation in scents with the function s.class, package ade4 v.1.7-22 (Thioulouse et al., 2018). The ratios of VOC emissions between the two species and the effect of appendix and air temperature were compared with the package Vegan v.2.5-7 (Oksanen et al., 2019) and the Adonis2 function by a PERMANOVA test (Bray–Curtis distance, n = 999 permutations). The relative abundance of chemical families was compared between species using a generalized linear mixed model (GLMM, function glmmTMB, glmmTMB package v.1.1-9; Brooks et al., 2024), with individual level as a random variable and a beta distribution.

Thermogenesis

For stamen thermogenesis, the effect of air temperature, species and peak number (P1, P2, P3st and P4) was analysed using a linear mixed model (LMM) test, with the individual level as a random variable. For appendix thermogenesis (P3), the effect of species and air temperature was tested by simple linear models. Given the high number of tests performed on thermogenesis dataset, we adjusted the P-value with the FDR function from the fuzzySim package (v.4.9.9; Barbosa, 2015) to correct the inflated type I error (Supplementary Data Table S2).

Activity

For the diel activity pattern, we tested the effect of the sex and light treatment on the counts per hour during daytime versus nighttime phases. A GLMM test, with glmmTMB function, was used with a negative binomial distribution, and the individual level was set as a random effect. The impact of the simplified scent on activity level of moth flies was tested with a general linear model with a quasi-Poisson family (Ver Hoef and Boveng, 2007).

RESULTS

Scent analysis

Globally, the flower scent differed between the two species (F_1,10 = 12.06, R² = 0.59, P = 0.003; Fig. 2). For A. italicum, the five major VOCs released by inflorescences during sampling represented ∼71 % of the floral scent composition and 75 % for A. maculatum (Fig. 2). The five major VOCs for A. italicum were the 3,7-dimethyl-oct-1-ene (mean ± s.e.: 23.1 ± 3.9 %), β-citronellene (21.5 ± 2.6 %), β-caryophyllene (10.1 ± 2.1 %), α-citronellene (8.2 ± 1.7 %) and an unknown compound (retention index = 1665; 8.3 ± 3.8 %). For A. maculatum, the five major VOCs were the indole (25.1 ± 6.1 %), an unknown compound (retention index = 1665; 21.0 ± 6.1 %), p-cresol (16.5 ± 6.5 %), 2-heptanone (7.0 ± 2.2 %) and α-humulene (5.0 ± 1.7 %). The relative proportions of chemical families of VOCs differed between species, with an interaction between species and chemical family (${\chi }_6^2=92.25$, P < 0.001; Supplementary Data Fig. S4). The most represented families were the monoterpenes (mean ± s.e.: 60.5 ± 7.9 %) and sesquiterpenes (20.3 ± 5.2 %) for A. italicum. Conversely, we observed a predominance of nitrogen-containing compounds (29.6 ± 5.6 %) and aromatic molecules (19.8 ± 6.9 %) for A. maculatum. Only the ratios of aliphatic compounds, monoterpenes and nitrogen-containing compounds differed significantly between the two species (emmeans’ test, P = 0.01, P < 0.0001 and P < 0.0001, respectively). For each species, a strong heterogeneity in the relative proportion of chemical families was observed at the inter-individual level.

Fig. 2.

Barplot of the mean relative proportion (as a percentage) of volatile organic compounds (VOCs) of A. italicum (blue; n = 5) and A. maculatum (red; n = 6). The names of the main VOCs (threshold of 5 %) are in bold. Top right inset, graphical representation of the redundancy analysis with the two Arum species. The variance explained by each axis is detailed.

Flower temperature patterns

Stamens

The peak temperatures of the stamens did not differ between the two Arum species (${\chi }_1^2=1.81$, P = 0.22) but according to the peak (P1, P2, P3st and P4; ${\chi }_3^2=41.37$, P < 0.001). The stamens reached higher temperatures during the P2 heat peak of thermogenesis [Supplementary Data Fig. S5; 27.4 ± 0.7 °C (mean ± s.e.)] compared with the other peaks of the stamen thermogenic sequence. Overall, the temperature of stamens was positively correlated with air temperature (Fig. 3A; ${\chi }_1^2=29.35$, P < 0.001). The excess temperature of the stamens did not differ between the two species (${\chi }_1^2=0.98$, P = 0.23) but decreased with increasing air temperature (${\chi }_1^2=8.92$, P < 0.001), suggesting an imperfect thermoregulation behaviour (Supplementary Data Fig. S6). The excess temperature of the stamens varied according to the heat peak (${\chi }_3^2=37.79$, P < 0.0001), with the highest excess temperature observed at P2 (Fig. 3B; 9.7 ± 0.7 °C). The excess temperatures at the other peaks did not differ from each other.

Fig. 3.

Thermogenesis of stamens. The blue line is the linear regression and the dotted line depicts the x = y isothermal line. (A) Temperature of stamens as a function of the air temperature (stamen thermogenesis peaks and both species were pooled; n = 83). (B) Excess temperature of arums at each peak of thermogenesis. The result of the post hoc test is notified by the letters. (C) Minimum temperature of stamens during the visit of pollinators as a function of the air temperature for both Arum species. (D) Minimum temperature of stamens according to the Arum species. The data points are represented by coloured circles, and error bars show s.e.m.

At that time of the season, active pollinators face the challenge of limiting their exposure to low temperatures. As expected, the minimum stamen temperature during the first night of anthesis increased with air temperature (Fig. 3C; F_1,20 = 231.01, P < 0.001). The minimum stamen temperature was warmer in A. italicum (Fig. 3D; F_1,20 = 6.88, P = 0.03). Similar results were observed for excess temperature, with an increase of the minimum stamen temperature excess when the air temperature increased (F_1,20 = 20.63, P < 0.001). The excess minimum temperature was more important for A. italicum (F_1,20 = 6.54, P = 0.03), with a mean of 4.7 ± 0.8 °C, compared with 2.6 ± 0.4 °C for A. maculatum.

Appendix

The second and warmest thermogenic part of the inflorescence, the appendix, is responsible for the scent emission. The appendix was proportionally warmer as air temperature increased (Fig. 4A; F_1,20 = 6.78, P = 0.03) and varied across species (Fig. 4B; F_1,20 = 26.32, P < 0.001), with the appendix of A. italicum being warmer (31.1 ± 1.2 °C) than that of A. maculatum (24.5 ± 1.2 °C). However, the volume of the appendix was about six times larger in A. italicum (Fig. 4C; F_1,20 = 44.86, P < 0.001). Thus, the appendix temperature normalized by its volume differed between the two species (Fig. 4D; F_1,19 = 21.04, P < 0.001), with an appendix heat accumulation about four times larger in A. maculatum (27.4 ± 3.9 °C cm⁻³) than in A. italicum (6.1 ± 0.9 °C cm⁻³). The air temperature had no effect on the appendix temperature normalized by its volume (F_1,19 = 0.03, P = 0.89).

Fig. 4.

Thermogenesis of appendix. (A) Relationship between the maximum temperature of the appendix (P3) for both Arum species and the air, and their associated linear regression. The dotted line depicts the x = y isothermal line. (B–D) Appendix maximum temperature (in degrees Celsius) (B), Appendix volume (in centimetres cubed) (C) and maximum temperature normalized by its volume (in degrees Celsius per centimetre cubed) (D) according to the Arum species during P3. The data points are represented by coloured circles, and error bars show s.e.m.

The appendix temperature excess and temperature excess normalized by its volume were not modified by the air temperature (F_1,20 = 3.62, P = 0.10 and F_1,19 = 3.35, P = 0.11, respectively). However, these two floral traits differed between the two Arum species. The temperature excess of the appendix was about twice as high in A. italicum (F_1,20 = 22.38, P < 0.001), with mean of 18.5 ± 1.4 °C, as in A. maculatum, with a mean of 8.6 ± 0.9 °C. The relationship was reversed when the temperature excess normalized by the volume was used, with an appendix heat accumulation more important in A. maculatum (Supplementary Data Fig. S7; F_1,20 = 26.57, P < 0.001).

Timing of thermogenesis

The timing of the three peaks of stamen thermogenesis (P1, P2 and P4) did not differ between the two Arum species (${\chi }_1^2=2.34$, P = 0.16) and it did not depend on air temperature (${\chi }_1^2=0.80$, P = 0.40). Moreover, these three stamen thermogenic peaks reached their maximum temperature at the same time during their respective day (${\chi }_2^2=2.24$, P = 0.36), at 14:00 ± 00:10 h (mean ± SE). In addition, the timing of P3 (following the opening of the spathe, i.e. anthesis) for A. maculatum tended to be earlier than that of A. italicum (18:00 ± 00:45 and 20:35 ± 00:55 h, respectively) but this was not statistically supported (F_1,20 = 3.94, P = 0.09). The air temperature did not affect the timing of flowering (F_1,20 = 0.0002, P = 0.98).

Relationship between temperature patterns and emission of VOCs

The air temperature and the appendix temperature had no effect on the scent composition of the flowers (P = 0.52 and P = 0.78, respectively). Likewise, the air and the appendix temperature did not influence the relative proportions of the respective major VOCs based on the threshold of 5 % (A. italicum: P = 0.16 and P = 0.52, respectively; A. maculatum: P = 0.19 and P = 0.17, respectively).

Pollinator activity

In controlled conditions (14 h light–10 h dark; first experiment), activity levels differed slightly between sexes, with males being more active than females (Fig. 5A; ${\chi }_1^2=5.15$, P = 0.023). The light conditions influenced the activity level of P. sigma (${\chi }_1^2=579.22$, P < 0.001). Psychoda sigma activity was almost 30 times higher during the scotophase (nighttime) than during the daytime (Fig. 5B). In the absence of external cues, when flies were exposed to constant light conditions, P. sigma still responded in a similar manner to the time periods corresponding otherwise to light/dark conditions during their rearing. Psychoda sigma was more active during the supposed nocturnal period [24 h night: ${\chi }_1^2=315.97$, P < 0.001 (Supplementary Data Fig. S8) and 24 h light: ${\chi }_1^2=118.97$, P < 0.001]. Therefore, global activity levels remained lower during perpetual day compared with perpetual night, reinforcing the conclusion that this species is nocturnal. Sex had no effect on activity in constant light treatments (P > 0.05).

Fig. 5.

Pollinator activity in controlled light conditions (14 h light–10 h dark). (A) Mean count of crossing of the infra-red barrier for males (blue line; n = 17) and females (pink line; n = 17) every 5 min during 24 h. Shaded areas correspond to the night. (B) Count per hour (pooling both sexes) during the dark and light periods in the same experiment (n = 34).

When P. sigma females were exposed to the 14 h light–10 h dark cycle, their activity level was modulated by the presence of floral scent representative of the two arums (second experiment) only in the night portion of the cycle (Fig. 6; dark: ${\chi }_3^2=18.58$, P < 0.001; light: ${\chi }_3^2=5.64$, P = 0.13; Supplementary Data Fig. S9). A higher number of counts was observed for females in the first hour of the night with the smell of horse dung in comparison to the solvent (contrast: P = 0.002). Then, the activity of female P. sigma did not differ between the simplified scent of A. maculatum and the smell of horse dung, and it was higher than the activity level with the solvent alone (P = 0.56 and P = 0.04, respectively). Flies in the presence of the simplified odour of A. italicum tended to be less active, but it did not differ from the smell of dung, the simplified scent of A. maculatum or the solvent (P = 0.05, P = 0.56 and P = 0.38, respectively).

Fig. 6.

Number of counts during the first hour of the night as a function of the scent conditions: solvent (pentane, n = 21), dung (horse dung, n = 22), simplified smell of A. italicum (n = 24) and simplified smell of A. maculatum (n = 22). The data points are represented by coloured circles, error bars show the s.e.m., and letters indicate the significant differences.

DISCUSSION

A few key studies have investigated pollination strategies with an integrative approach (Terry et al., 2007), highlighting the need for linking the mechanisms behind emission of VOCs, floral physical properties and insect pollinator behaviour. Until now, scent and thermogenesis of temperate arums have often been treated separately, with few attempts to relate the two (Marotz-Clausen et al., 2018). In this study, we confirm that the two Arum species (A. italicum and A. maculatum) display distinct scent (Chartier et al., 2013) when growing in sympatry. The distribution ranges of both species overlap mainly in Western Europe for a few weeks in spring. Although this sympatry could promote convergent floral trait selection by pollinators, the short adult lifespan (2 days at 20 °C) and deceptive pollination might promote or constrain this convergence. We further demonstrate that these two arums have, for the appendix emitting scent, slightly different thermogenic strategies. Although A. italicum reaches higher temperatures, the thermogenic process in A. maculatum seems more efficient, with a higher temperature when normalized by appendix volume. However, the floral chemistry remains robust to variations in flower temperature and air temperature. Finally, our experiments confirmed that one of their pollinators is nocturnal and that A. maculatum emits a more stimulating scent. This might serve to compensate for its weaker thermogenesis and/or to enhance (earlier) pollinator attraction from the very start of the insect active period (sunset). Taken together, our results illustrate the role of fine-tuned processes in volatile emission, thermogenesis and insect behaviour in the pollination ecology of the two sympatric plants.

Our chemical analysis confirmed previous findings on the scent divergence between A. maculatum and A. italicum (Chartier et al., 2013, 2016). This scent divergence partly explains the contrasting assemblages of pollinators found in other studies, with A. maculatum being more specialist of Psychoda visitors than A. italicum (Albre et al., 2003; Espíndola et al., 2011). However, an unknown compound (retention index = 1665) was found in high proportion for both species. We cannot exclude the possibility that this unidentified compound explains the ability of A. italicum and A. maculatum to attract the same genus of pollinators (Psychoda) in large proportion (Chartier et al., 2013). Nevertheless, our reduced list of VOCs (threshold of 1 % of relative proportion) is certainly relevant in the process of pollinator recruitment because it contains the most represented compounds. Future studies should confirm the extent to which the presence of minor compounds plays a role in pollinator attraction. Electroantennographic detection on Psychoda antennae showed that not only the most abundant VOCs of A. maculatum elicited a neurophysiological response, but also numerous minor compounds (i.e. 1-octen-3-ol, 2,6-dimethyl-7-octen-2-ol) (Gfrerer et al., 2022).

Nevertheless, the simplified odour of arums stimulated the activity level of females P. sigma, but only during activity window (nocturnal) of the insect. This result suggests that scent emission before sunset can become counterproductive for the plant, which cannot stimulate the activity levels of P. sigma at this time of the day. Flowering after sunset probably increases the likelihood of inflorescences being visited. Finally, we highlighted that females of P. sigma were equally well stimulated by A. maculatum simplified floral odour as by the odour of horse dung, whereas the results concerning the simplified floral odour of A. italicum were less clear. This could be a first explanation for the greater specialization of A. maculatum for pollinators of the genus Psychoda (Gfrerer et al., 2022). Although we hypothesize that the stimulation of activity is necessary to decrease the time needed to ‘trap’ pollinators, additional experiments are needed to quantify the attractant effect on other pollinator species to gain a global view of these pollination processes. Moreover, the composition of VOCs seems to be resilient to variations in both flower and air temperature. Indeed, we showed recently that the scent of A. italicum is particularly robust to environmental stressors, such as warmer soil and exposure to solar radiation (Leclerc et al., 2024). The attractive system of arums might be robust to environmental change, but the responses of Psychoda flies to these same environmental perturbations remain unknown. The second floral trait that matters for pollination ecology is thermogenesis, and some differences were found between the two species. We highlighted a strong difference in appendix temperature in favour of A. italicum. The warmer appendix could maximize odour diffusion and act as a ‘candle in the night’ if pollinators are equipped with thermal receptors, although this remains to be demonstrated. Nevertheless, we showed that this organ, although slightly sensitive to air temperature, remains an almost perfect thermoregulator, tending towards a temperature optimum (∼30 °C) for scent emission and/or enzyme activity (Ito et al., 2020), suggesting temperature control as shown in A. italicum (Leclerc et al., 2024). Thermogenesis of the second organ (stamens) did not differ between the two species and showed a daily pattern of thermogenesis around 14:00 h for 3 days consecutively. However, the temperature of each peak was not similar. The warmer P2 preceding the spathe opening would not be directly linked to pollinator attraction but rather to flower maturation and spathe unfolding (Albre et al., 2003). In addition, P4 takes place during the presence of pollinators inside the floral chamber when the pollen is released. It would be a physiological constraint for the pollen maturation (Albre et al., 2003) and also for pollinator survival relative to its low thermal tolerance (Leclerc et al., 2022); this dual constraint might leave little room for interspecific differences. Future research should investigate the functional role between the floral chamber microclimate and the ecophysiological traits of its pollinators.

Although the effect of flower temperatures on pollinators is often associated with warm temperatures (Harrap et al., 2017), the importance of cool temperatures should not be overlooked (Chown and Nicolson, 2004; Colinet et al., 2015). We showed that the stamen temperatures during the night when pollinators are trapped are always above air temperature and were warmer in A. italicum than in A. maculatum. Therefore, Psychoda and other pollinators could benefit from the warmer temperature in the inflorescence in A. italicum, like other insects that find thermal refuges during winter (Danks, 1978; Alford et al., 2014; Reid and Harley, 2021). More research is needed to confirm whether a mechanism with heat reward exists for pollinators in this temperate Arum system, as in another tropical Araceae (Seymour et al., 2003), thereby calling into question the extent to which this system is deceptive.

Beyond the scent composition and temperature, the phenology of the different thermogenesis peaks is crucial to ensure pollination. We highlighted the robustness of the timing of all the thermogenesis peaks, although we noted an earlier trend in the onset of the appendix peak in A. maculatum. This finding potentially shows an advantage for A. maculatum on the attraction of pollinators in sympatric (competitive) areas, because it would be the first of the two arums to attract and trap pollinators. However, this earlier flowering is also constrained by the nocturnal activity of P. sigma, one of their main pollinators, and could become detrimental for A. maculatum as sunset occurs later as the season progresses (Leclerc et al., 2024). Thus, a more important mismatch between the floral signal (well before sunset) and the period of occurrence of pollinators would lead to a significant loss of fitness in a similar way to a climate change-induced phenological mismatch occurring at seasonal scale (Renner and Zohner, 2018).

To conclude, the two sympatric arums display subtle divergences in their deceptive strategy to attract pollinators. Arum maculatum is limited by its appendix size and displays a weaker thermogenic peak during scent emission. This might be compensated by: (1) emitting the scent slightly earlier than A. italicum to ensure it attracts the pollinators first; and (2) increasing the pollinator activity to a higher level than that of A. italicum. These ecophysiological differences between the two arums suggest that selection pressures might apply from competing interference for the same pollinators. Furthermore, this type of competition takes place in a larger context than our biological system. In other angiosperms, a strong density of co-flowering plants with the same pollinators can decrease the rate of pollinator visits (Kehrberger and Holzschuh, 2019). However, the lack of knowledge on the evolutionary divergence of floral traits in co-flowering plants (except orchids) remains limiting.

SUPPLEMENTARY DATA

Supplementary data are available at Annals of Botany online and consist of the following.

Table S1: list of volatile organic compounds used in chemical analyses. Table S2: statistical outputs of the linear mixed-effects models and linear models showing raw and adjusted P-values with the FDR function. Figure S1: mass spectra of the four unknown volatile organic compounds. Figure S2: schematic diagram of the thermogenesis cycle of arums. Figure S3: illustrations of the activity recording system. Figure S4: proportion of chemical families (as a percentage) for the individuals of both Arum species. Figure S5: temperature of stamens as a function of the heat peak across the thermogenesis cycle. Figure S6: relationship between the excess stamen temperature and air temperature. Figure S7: excess appendix temperature normalized by its volume according to the Arum species. Figure S8: counts per hour during the theoretical dark and light periods under constant darkness. Figure S9: number of counts during the last 15 min of the light period as a function of the scent condition. Supplementary Results: minor supplementary results concerning the duration and the maximum excess of temperature.

FUNDING

This work was supported by a PhD fellowship from the University of Tours to M.L. and the grant HOTPALM provided by the French National Agency for Research [ANR 20-CE32-0003-01].

AUTHOR CONTRIBUTIONS

Conceptualization and experimental design: M.L., M.G. and S.P. Experiments and monitoring: M.L. Chemical analyses: M.L., M.G. and E.P. Data and statistical analyses: M.L. and S.P. Original draft writing and preparation: M.L. and S.P. Review and final editing: M.L., M.G., E.P. and S.P. The authors declare no conflict of interest.

DATA AVAILABILITY

Data and code are archived in Zenodo: https://doi.org/10.5281/zenodo.8289611.