Thermogenesis‐derived spatiotemporal microclimates guide pollinator movement to ensure pollination

ABSTRACT

Thermogenesis represents one of the oldest pollination traits; yet, the ecological function of the microclimates that it creates remains poorly understood. Here, we elucidate how the spatiotemporally dynamic microclimate of Alocasia odora orchestrates pollinator movement within its inflorescence. Upregulation of alternative oxidase (AOX) in the upper spadix generates a steep, diurnally fluctuating thermal gradient against the non‐thermogenic lower chamber, forming a “thermal map”. We demonstrate that this thermal push–pull mechanism, interacting synergistically with the insects' innate requirements, drives the cyclical vertical movements of its specialized Colocasiomyia pollinators. While strict feeding requirements establish a baseline preference for the lower sterile zone, species‐specific oviposition preferences dictate their distinct movement trajectories. Specifically, C. xenalocasiae primarily drives pollen importation, whereas C. alocasiae acts as the dominant contributor to pollen exportation. In contrast, floral scent primarily functions in long‐distance pollinator attraction, while light plays only a minor modulatory role in this fine‐scale within‐inflorescence navigation. Our comparative analysis reveals a prevalent spatiotemporal separation of sexual organs across thermogenic flowers globally. We therefore propose that microclimate‐guided pollinator movement—where thermal gradients overcome structural constraints to ensure reproductive success—constitutes a widespread, yet underappreciated evolutionary strategy.

Alocasia odora's microclimatic heterogeneity guides Colocasiomyia fly movement within its inflorescence. During the female stage, midday heat in the upper spadix pushes flies down to the cooler female zone. During the male stage, nocturnal warmth in the upper part pulls flies upward and spathe constriction closes the lower female chamber.

INTRODUCTION

In mutualistic plant–pollinator relationships, plants utilize various cues—including visual specialty, scent, humidity, and even temperature—to influence pollinator behavior and maximize their reproductive success (Hempel de Ibarra et al., 2015; Harrap et al., 2017; Lawson et al., 2018; Salzman et al., 2023). Among these, endogenous floral heating (thermogenesis) is an ancient trait (Watling et al., 2008; Barreto et al., 2025). It has been shown to enhance scent volatilization for pollinator attraction (Terry et al., 2004, 2016; Wang et al., 2022), provide warmth as a thermal reward (Seymour et al., 2003), and even generate heat radiation as a direct signal to attract visitors (Valencia‐Montoya et al., 2025). More importantly, thermogenesis creates microclimates (localized climatic conditions) within or around the inflorescence that differ from the macroenvironment (Norgate et al., 2010; van der Kooi et al., 2019; Apland and Koski, 2025). These microclimates can be crucial for ectothermic insect pollinators (Herrera, 1995; McCallum et al., 2013), which must seek suitable thermal environments to regulate their body temperature (Dyer et al., 2006; Shrestha et al., 2018; Dahake et al., 2025). Some thermogenic flowers can develop distinct internal microclimates. For example, in the Araceae—the family with the greatest diversity of thermogenic species—many flowers form complex spadix structures where only specific parts produce heat, while others remain non‐thermogenic (Angioy et al., 2004; Ivancic et al., 2008). Consequently, thermogenesis creates a spatially heterogeneous microenvironment within these inflorescences, and this heterogeneity may follow a diurnal temporal dynamic (Roemer et al., 2022). However, the effects of such floral thermal microclimates on pollinator behavior and their role in pollination within an inflorescence remain unclear (Angioy et al., 2004; Harrap et al., 2020).

We studied the brood‐site pollination system between Alocasia odora (Araceae) and its specialized pollinators, Colocasiomyia flies (Drosophilidae) (Yafuso, 1994; Toda et al., 2022). The aroid is protogynous (the female stage precedes the male stage) and self‐incompatible; its female (receptive) and male (pollen‐releasing) stages each last 1–3 d (Ding et al., 2024). The spadix of A. odora is surrounded by a modified bract termed the spathe (Figure 1A–D). This spathe constricts midway to divide the spadix into two discrete upper and lower chambers. The upper chamber houses the non‐reproductive appendix and staminate floret zone, whereas the lower chamber contains the sterile zone and the pistillate floret zone (Yafuso, 1993). This spatial organization—typical of many monoecious Araceae with unisexual flowers—imposes a temporally staged, spatially opposed sequence for successful pollination (Yafuso, 1993). Specifically, pollen‐carrying visitors that initially land on the upper inflorescence move basipetally (downward) to the receptive female zone to deposit outcross pollen. Subsequently, during the male phase, they move acropetally (upward) to the male zone to pick up pollen for subsequent dispersal before visiting the next flowers (Bertin and Newman, 1993; Chartier et al., 2014; Jiménez et al., 2019).

Figure 1.

Pollinator movement between floral zones promotes pollen import and export in Alocasia odora

(A) An A. odora individual shows consecutive flowering, with flower buds, an inflorescence in bloom, several developing fruits, and fallen fruits present simultaneously. (B) During the female stage, diurnal activities of Colocasiomyia flies on the sterile and female zones in the lower part of the inflorescence facilitate pollen import. For observational purposes, portions of the spathe were excised to expose the internal structure of the spadix. (C) During the female stage, nocturnal activities of Colocasiomyia flies occur on the male zone in the upper part of the inflorescence. (D) During the male stage, spathe constriction prevents flies from re‐entering the lower part, promoting pollen export as flies depart toward newly opened, receptive inflorescences. White scale bar: 1 cm.

Complicating this navigational requirement is the behavioral ecology of the pollinators. Two Colocasiomyia fly species, C. alocasiae and C. xenalocasiae, co‐occur on the inflorescences of A. odora and depend entirely on them for feeding, mating, and oviposition (Bian et al., 2021; Toda et al., 2022). Pollinators utilize these floral regions in a stereotyped manner: while both species feed primarily in the lower chamber (sterile and female zones), they show marked differences in oviposition‐site preferences. C. xenalocasiae females deposit eggs exclusively in or near the female zone, whereas C. alocasiae females preferentially oviposit in the male zone (Yafuso, 1994). These divergent innate requirements inherently shape their baseline spatial distributions within the inflorescence. Therefore, understanding how dynamic floral thermogenesis interacts with these species‐specific life‐history traits to orchestrate the precise temporal movements required for successful pollen transfer is essential for a comprehensive view of this specialized pollination system.

In this study, we demonstrate that in Alocasia odora, thermogenesis creates microclimatic heterogeneity within the inflorescence, which in turn guides the regular diel movement of its pollinators. This mechanism facilitates pollinator activity in specific floral zones and promotes the plant's pollination success. Specifically, during the day, the warm microclimate in the thermogenic upper part (appendix and male zone) causes Colocasiomyia flies to avoid this area, whereas at night, as ambient temperatures drop, the heat in this region attracts them. This diurnal reversal of thermal gradients establishes a “push–pull” pollination mechanism within a single inflorescence. Importantly, this thermally driven movement interacts with the distinct oviposition requirements of the two pollinator species, resulting in different contributions to the male and female functions of pollination. Furthermore, by reviewing and comparing thermogenic traits across flowers from 15 families, we propose that this pattern—where thermogenesis promotes pollinator movement within and between inflorescences—is likely a widespread, yet long‐overlooked phenomenon.

RESULTS AND DISCUSSION

Effects of feeding and oviposition requirements on pollinator movement

The entire life cycle of the specialized pollinators Colocasiomyia alocasiae and C. xenalocasiae occurs exclusively on the inflorescences of Alocasia odora. Consequently, the various innate requirements of these pollinators—such as adult feeding and female oviposition—significantly influence their microhabitat selection and movement patterns among floral zones. Nutritionally, during the female stage of the inflorescence, no pollen is available for the flies to consume. Although pollen becomes available during the subsequent male stage, the sponging mouthparts of Colocasiomyia flies render them incapable of directly ingesting solid particles such as pollen grains (Coronado‐Gonzalez et al., 2008). Consistent with this dietary restriction, our results indicate that over 60% of adult flies of both species (including both sexes) aggregated in the sterile zone, where they showed distinct feeding behaviors (Figure 2A). Notably, at the base of the sterile zone adjacent to the female zone, most flies remained for extended periods, displaying frequent lapping movements of their mouthparts. Specifically, 86.6% of C. alocasiae males and 92.1% of C. xenalocasiae males were highly aggregated in the sterile zone (Figure 2B, C). The spatial distribution of male flies serves as a direct indicator of their intrinsic foraging requirements, as it eliminates confounding factors associated with the presence of conspecific females, particularly mating motivation. These results confirm that the sterile zone serves as the sole food resource and a critical feeding site for adult males of both species.

Figure 2.

Feeding and oviposition preferences of Colocasiomyia flies on Alocasia odora inflorescences

(A) Schematic illustration of the laboratory behavioral assay setup, where sectioned spadices (male, sterile, and female zones) were placed in tubes to observe fly feeding and spatial distribution. (B, C) Spatial distribution of male (black dots) and female (red dots) adult (B) C. alocasiae and (C) C. xenalocasiae across the functional floral zones at 1 h post‐introduction. Data are expressed as the proportion of flies in each respective zone. (D) Representative photographs and schematic showing the distinct oviposition microhabitats of the two species. The red arrow/border indicates C. alocasiae eggs inserted into the narrow slits of the floral structures, and the teal arrow/border indicates C. xenalocasiae eggs deposited in the female zone. White scale bar: 5 mm. (E, F) Quantification of oviposition site preferences for C. alocasiae (orange dots) and C. xenalocasiae (teal dots) across different inflorescence regions in (E) controlled laboratory assays (using sectioned spadices) and (F) natural field experiments (using intact, netted inflorescences) over a 24‐h period. Individual data points are shown, with error bars representing the means ± SEM.

In contrast to males, females of the two species showed distinct distribution patterns. Specifically, C. xenalocasiae females (91.2%) also aggregated primarily in the sterile zone, suggesting that their feeding requirements spatially align with their oviposition preferences (Figure 2C). However, C. alocasiae females displayed a more dispersed distribution: 65.1% were found in the sterile zone, while 28.3% were located in the male zone (Figure 2B). This suggests that the spatial distribution of C. alocasiae females is strongly influenced by their unique oviposition requirements. To uncouple the effects of feeding and oviposition requirements, we examined the oviposition site preferences of both fly species in the absence of interspecific interference. The results revealed striking differences in oviposition site preference patterns between C. alocasiae and C. xenalocasiae (Figure 2D–F). Notably, C. alocasiae females primarily selected either the upper male zone or the sterile zone for oviposition, inserting most of their eggs into narrow slits of the floral structures (Figure 2D). They showed a strong preference for sites that develop into the post‐anthesis male zone, as these slits are precisely matched to the width of their ovipositors (unpublished data). Conversely, C. xenalocasiae females deposited all their eggs in the lower female zone or on the adjacent spathe surface, a pattern consistent across both laboratory assays and field experiments (Figure 2D–F). These egg distribution patterns align with observations from other A. odora populations, such as populations in Japan and China, indicating that oviposition site preference is a relatively conserved trait at the species level (Toda et al., 2022).

Consequently, the feeding requirements of both Colocasiomyia species drive them to move downward from their initial landing sites in the upper part of the inflorescence toward the lower sterile zone in the female stage. However, from the perspective of oviposition behavior, C. xenalocasiae females prefer to lay eggs in or near the female zone, meaning that their oviposition sites substantially overlap with their feeding sites in the sterile zone. Thus, oviposition requirements likely further reinforce their restriction to the lower part of the inflorescence. In contrast, the preference of C. alocasiae females to oviposit in either the male zone or the sterile zone prompts them to maintain a higher level of activity in the upper part of the inflorescence during the male stage. Taken together, feeding and oviposition requirements shape distinct, species‐specific directional movements within the inflorescence: females of both species move downward during the female stage of A. odora, whereas C. alocasiae females shift upward during the male stage. However, these stage‐dependent movement trajectories alone are insufficient to explain the rhythmic bidirectional movements repeatedly observed in these two pollinator species.

Thermal microclimates guide pollinator movement within the inflorescence

We first confirmed the differential thermogenic activity in the inflorescence of Alocasia odora and further elucidated its underlying molecular basis. Although the general pattern of heat production, which is mainly confined to the upper inflorescence (appendix and male zone), has been reported previously, the fine regulatory details of this thermogenesis remain unclear (Bian et al., 2021). We therefore localized and inferred the molecular basis of active heat production by focusing on the expression of alternative oxidase (AOX) and uncoupling proteins (UCPs), which are key mediators of respiratory heat generation (Ito, 1999; Zhu et al., 2011; McDonald, 2023). A de novo transcriptome assembly yielded 85,898 unigenes (N50 = 1.85 kb), with homology searches revealing three AOX (AOX1a, AOX1b, and AOX4) and three UCP (UCP1, UCP5, and UCPB) homologs.

Differential expression analysis based on raw read counts (DESeq. 2) identified AOX1b as the key candidate gene consistently and specifically upregulated in the thermogenic tissue (male zone during the female stage) compared to both non‐thermogenic controls (Figure S1). Specifically, AOX1b transcript levels in the male zone were significantly higher than those in the corresponding male zone of bud‐stage inflorescences (Log₂(Fold change) = 4.33, adjusted P = 4.75 × 10⁻⁸) and the adjacent female zone (Log₂(Fold change) = 10.84, adjusted P = 1.59 × 10⁻²⁸). In contrast, other AOX and UCP isoforms did not show consistent significant enrichment across these comparisons (e.g., UCPB was not significantly upregulated at female stage vs. the bud stage, adjusted P = 0.58) (Table S2). Previous studies indicate a substrate specificity between these pathways: Plant thermogenesis via AOX primarily consumes carbohydrates, whereas thermogenesis mediated by UCP predominantly utilizes lipids (Ito and Seymour, 2005; Valencia‐Montoya et al., 2025). In the thermogenic regions of the A. odora inflorescence, our data demonstrate that the male zone operates via AOX. Taken together, the transcriptome expression signature and thermal imaging support a model of localized activation of the alternative oxidase pathway in the upper part (appendix and male zone), whereas the lower part (sterile and female zones) shows little to no thermogenic activity (Bian et al., 2021).

Building on this spatial temperature gradient, our continuous monitoring throughout anthesis revealed that the heterogeneous thermal microclimates within the inflorescence are closely synchronized with fly distributions along the spadix. The heat production in the upper inflorescence fluctuated with ambient temperature rather than holding a constant temperature: On average, the appendix and the male zone were 2.4°C ± 1.1°C and 3.5°C ± 1.8°C above ambient, respectively (Figure 3A), in contrast to thermoregulatory species such as Nelumbo nucifera, which maintain 30°C–35°C floral temperatures (Seymour and Schultze‐Motel, 1996; Seymour, 2010; Wang and Zhang, 2015). The upper–lower temperature difference (ΔT) correlated closely in space and time with changes in fly abundance. We propose that fly activity is largely confined to a “thermal preference range” of approximately 20°C–25°C, which aligns with the preferred temperature range of many widespread drosophilid species (MacLean et al., 2019; Ito and Awasaki, 2022). Future targeted temperature gradient choice assays will be required to validate this range for the two Colocasiomyia species in this study. Relative to the diurnal ambient temperature, extremes above or below this range likely guide Colocasiomyia flies to seek more suitable thermal zones within the inflorescence (Figure 3B).

Figure 3.

Thermogenesis‐mediated temperature changes in the inflorescence correspond to the movement patterns of flies

(A) Typical temperature trajectories of different Alocasia odora spadix regions and ambient air over time. Red dashed lines mark the 20°C–25°C range, presumed to represent the preferred temperature range of Colocasiomyia flies. (B) Proportion of flies distributed in the upper parts (male + appendix) versus the lower parts (sterile + female) of the spadix (left Y‐axis), and the mean number of flies per inflorescence (right Y‐axis) during the same observation period, as shown in (A) (n = 3 inflorescences). For observation purposes, we widened the constricted area between the upper and lower parts of the spathe, allowing flies to access the lower part of the inflorescence even during the male stage. Note that either the entire spathe or a portion of the spathe in the illustration has been removed.

During the female stage, this preferred temperature zone shifted spatially—flies occupied the cooler, lower part during the day and moved to the warmer, upper part at night. During the male stage, thermogenesis in the upper part (male zone) persisted, albeit at reduced intensity compared to the female stage (Days 2–3) (Figure 3A). Flies were observed primarily in the upper part at night during this period (Day 3) (Figure 3B). Under natural conditions, the spathe constriction prevents re‐entry into the lower chamber, but sustained thermogenesis ensures that departing flies carry pollen from the male zone. Notably, for observation purposes, we had widened the constricted area of the spathe during the female stage, which allowed flies to continue accessing the lower chamber even during the male stage. This manipulation revealed that flies still attempted to enter the cooler lower part, suggesting that they remain thermally repelled by the upper zone during this period. Moreover, the total number of flies per inflorescence showed limited fluctuation across observations (means ± SD = 21.4 ± 7.2 individuals per inflorescence; n = 3 inflorescences monitored at hourly intervals over ~3 d), indicating that the observed distribution shifts reflected mainly within‐inflorescence relocation rather than exchange with other inflorescences.

To determine whether temperature microclimates directly influence flies' movement within the inflorescence and to assess the potential role of light (Terry et al., 2014), we conducted in situ microclimate manipulation experiments. At noon, cooling the upper part to < 25°C and heating the lower part to > 25°C prompted a rapid upward relocation, with the proportion of flies in the upper part increasing from 10.4% ± 6.9% to 89.2% ± 10.3% (means ± SD; post hoc Tukey's multiple comparisons, P < 0.05; n = 10) (Figure 4A). Conversely, at midnight, heating the lower part to > 20°C and cooling the upper part to < 20°C drove flies downward, reducing the fraction in the upper part from 91.4% ± 6.7% to 8.0% ± 9.9% (P < 0.05; n = 10) (Figure 4A). Integrating these manipulations with our behavioral observations, we propose a novel push–pull thermal mechanism: During the day, increasing heat in the upper part pushes flies toward the cooler lower part; at night, sustained thermogenesis keeps the upper part relatively warm, while the lower part cools to the ambient conditions, pulling flies back upward. A similar push–pull concept has been invoked to explain the short‐term expulsion of pollinators from male cones to female cones in dioecious cycads, mainly driven by odor (Terry et al., 2007). However, that phenomenon operates between individuals and appears to be less precise than the within‐spadix navigation documented here. In parallel, daytime shading induced partial relocation (upper‐part fraction from 12.1% ± 11.2% to 39.0% ± 25.9%; two‐tailed t‐test, P < 0.05, n = 13), whereas nighttime light supplementation had no detectable effect (from 79.8% ± 14.2% to 72.4% ± 25.0%; P > 0.05, n = 6) (Figure 4B). Together, these results show that a temperature gradient created by spatially structured thermogenesis orchestrates pollinator movement within the inflorescence, establishing a fine‐scale microclimatic heterogeneity that serves as the dominant driver of navigation within the spadix (Caillon et al., 2014), with light serving at most as a secondary modulator. We note that both the absolute temperature and the temperature difference (ΔT) between floral zones likely contribute to guiding fly movement. Our continuous data show that the upper–lower ΔT fluctuated over the monitoring period and was temporally associated with distribution shifts (Figure 3B). While the precise sensory mechanism—whether flies respond primarily to local absolute temperature or to spatial thermal contrast—remains to be determined, drosophilids are known to detect both, and the combined cues likely enhance the robustness of within‐inflorescence navigation (Alpert et al., 2022; Evans et al., 2023).

Figure 4.

Temperature and light influence the movement of Colocasiomyia flies within the inflorescence

(A) Proportion of flies in different parts after experimental manipulation of regional temperatures during the day and night, respectively (n = 10 inflorescences). Statistical analysis was performed using a one‐way ANOVA, followed by Tukey's multiple comparisons. (B) Proportion of fly distribution under manipulated light conditions in different parts during the day (shading, n = 13 inflorescences) and night (illumination, n = 6 inflorescences), respectively. Statistical analysis was performed using a two‐tailed t‐test. *P < 0.05, **P < 0.01, and “ns” indicates no significant difference.

Floral scent composition is not the key factor driving pollinator movement within the inflorescence

In cycads, the push–pull pollination mechanism between male and female cones is considered to be primarily mediated by volatiles (Terry et al., 2007; Salzman et al., 2020). We therefore asked whether the diel movement of Colocasiomyia flies within the Alocasia odora inflorescence is similarly influenced by odor differences among floral regions. We observed that flies predominantly occupied the lower part of the inflorescence (sterile zone + female zone + lower spathe) during the day and the upper part (appendix + male zone + upper spathe) at night (Figure 5A). Their distributions differed significantly between noon and midnight (two‐tailed t‐test, P < 0.001; n = 23 inflorescences at noon, 16 at midnight; some inflorescences closed or entered the male stage by midnight, preventing accurate counting). While volatiles from the upper inflorescence have been implicated in attracting Colocasiomyia flies (Miyake and Yafuso, 2003), we first aimed to pinpoint the specific structure responsible for attraction within the inflorescence (female stage). Field four‐choice assays showed that only the appendix consistently attracted flies from a distance, with a mean capture of 4.4 ± 5.6 individuals (means ± SD; post hoc Dunn's multiple comparisons: P < 0.001, n = 30), whereas the male zone sporadically attracted only a few individuals (0.7 ± 1.3), and neither the sterile zone (0.03 ± 0.2) nor the female zone (0.06 ± 0.3) showed any meaningful attraction (Figure 5C). Paired close‐range assays further confirmed a significant preference for the appendix compared to the control, while no other floral region elicited a statistically relevant response (Wilcoxon signed‐rank test: appendix, P = 0.0015, n = 15; Figure 5D). Notably, in these close‐range two‐choice assays, the odor of the appendix became intensely pungent around the trap due to high local concentration; yet, it remained the only region that significantly attracted flies. This observation does not support the idea that high‐concentration appendix volatiles act as a repellent (“push”) to drive flies away from the upper inflorescence (Suinyuy et al., 2013). These findings indicate that the appendix emits either unique odorants or higher concentrations of key volatiles compared to other floral regions, especially the non‐thermogenic female zone.

Figure 5.

Floral scent is not the key factor driving the diel movement of pollinators

(A) Distribution of Colocasiomyia flies on Alocasia odora inflorescences at 12:00 (noon) and 24:00 (midnight) during the female stage (n = 16–23 inflorescences). Statistical analysis was performed using a two‐tailed t‐test; P < 0.001. (B) Dynamic headspace volatiles collected from the same inflorescences and ambient air at different times (n = 3–6 trials). Different letters indicate significant differences (P < 0.05), as determined by the Kruskal–Wallis test with Dunn's multiple comparisons. (C) Field four‐choice assay showing the number of Colocasiomyia flies captured by each floral region. Different letters indicate significant differences, as determined by a Friedman test, followed by Dunn's multiple comparisons (n = 30 trials). (D) Laboratory two‐choice assays and fly preference indices between each floral region and an empty control. **P < 0.01 and “ns” indicates no significant difference, as determined by the Wilcoxon signed‐rank test (n = 15 trials). (E) Volatile compounds emitted by the entire inflorescence and by each of the four separated regions (n = 2 trials). Compounds labeled in red are markedly enriched in the appendix.

To distinguish the composition and relative concentrations of volatiles from different inflorescence parts, we dissected spadices (with spathes removed) collected on the first day of the female stage into four regions: appendix, male zone, sterile zone, and female zone. Using solid‐phase microextraction coupled with gas chromatography–mass spectrometry (SPME–GC–MS), we found that the appendix had the most diverse and abundant bouquet of volatile organic compounds (VOCs) and best reflected the overall scent profile of the inflorescence (Figure 5E; Table S1). Among nine major inflorescence volatiles accounting for 94.5% of total emission, six were markedly enriched in the appendix: methyl propionate, methyl butyrate, methyl 2‐methylbutyrate, methyl benzoate, (E)‐4,8‐dimethylnona‐1,3,7‐triene, and methyl salicylate. In contrast, three volatiles showed little concentration difference across four regions: methyl isobutyrate, β‐caryophyllene, and humulene. Notably, most of these VOCs (methyl benzoate, (E)‐4,8‐dimethylnona‐1,3,7‐triene, methyl salicylate, β‐caryophyllene, and humulene) are also present in the closely related species Alocasia cucullata and are associated with attraction of the same pollinators (Miyake and Yafuso, 2005). Moreover, none of these compounds was detected in ambient air samples collected near the flowers, confirming that the detected signals originated from the inflorescence and not from environmental contamination. In gas chromatography–electroantennographic detection (GC–EAD) assays, female Colocasiomyia antennae responded to most appendix‐characteristic VOCs (except for methyl 2‐methylbutyrate) (Figure S2). Although antennal responses do not directly equate to behavioral attraction, they support the potential attractiveness of these major scent compounds.

Since the attractive scent emission primarily originates from the appendix, we performed continuous headspace collections from intact inflorescences at different times after anthesis to profile the dynamics of total scent emission, which serves as a proxy for emissions from the upper part. The results showed that the inflorescence emitted the highest volatile concentration at noon on the first day of the female stage, which subsequently decreased to ambient‐air levels by that same night. This was followed by a slight recovery at noon on the following day and again at noon on the third day when the inflorescence re‐entered the female stage (post hoc Dunn's multiple comparisons: Day 1 noon ≥ Day 2 noon = Day 3 noon ≥ Day 1 midnight = ambient; P < 0.05, n = 10) (Figures 5B, S3). This temporal pattern further indicates that floral scent is unlikely to be the key factor driving the diel vertical movement of Colocasiomyia flies within the inflorescence. If scent were the primary attractant, the upper part (appendix + male zone), which had the most diverse and abundant volatile bouquet at noon, would be expected to attract flies; however, flies were instead aggregated in the non‐attractive lower part (sterile zone + female zone) at that time. Conversely, at night, when volatile richness and abundance declined to near‐ambient levels, flies returned to the upper part. A limitation of our study is that we did not perform absolute quantification of volatile concentrations or emission rates from different floral zones, nor did we test how temperature fluctuations affect these parameters. It is possible that temperature modulates the release of specific volatiles, which could in theory contribute to the push–pull effect.

In summary, while a role for scent in the diel movement of pollinators cannot be entirely ruled out, our results support the conclusion that it is not the key driver. This conclusion is based on two lines of reasoning that challenge potential scent‐mediated mechanisms. First, from a perspective of scent attraction alone, the high‐concentration volatiles at noon did not demonstrate an attractive function, as pollinators primarily gathered in the lower part of the inflorescence where scent was weaker. Second, a scent‐based “push–pull” mechanism analogous to that in cycads is also unsupported (Terry et al., 2007). Our close‐range assays did not reveal a repellent effect of the higher relative abundance of key volatiles in the appendix; on the contrary, they showed significant attraction. Furthermore, no reduction in a putative repellent effect was observed during the male stage when the volatile profile had weakened. Therefore, we posit that the scent of the appendix primarily functions in long‐distance attraction, rather than in fine‐scale navigation within the inflorescence (Miyake and Yafuso, 2003).

Thermal microclimates drive pollinator movements to ensure efficient pollination in A. odora

During the female stage of anthesis in A. odora, adults of both Colocasiomyia species (C. alocasiae and C. xenalocasiae) move downward during the daytime from the thermogenic upper part of the inflorescence to the cooler sterile and female zones, where they feed on sugar‐ and amino acid‐rich exudates seeping from the surface of the sterile zone (Toda et al., 2022). If the flies that have arrived already carry pollen on their bodies, they will facilitate pollen import and pollination as they move downward. Notably, C. xenalocasiae females oviposit in the female zone, while C. alocasiae can oviposit in the sterile zone; thus, the C. alocasiae females have a lower probability of reaching the female zone. Moreover, the distinct spatial preferences driven by the oviposition requirements of Colocasiomyia females result in differential pollination efficiencies, suggesting that C. xenalocasiae likely acts as a greater contributor to pollen importation than C. alocasiae. At night, however, the adults move back upward from the lower part of the inflorescence to the appendix and the male zone in search of warmer microhabitats. Because the female stage—the receptive window during which stigmas remain viable for pollination—typically lasts 1–3 d for a given A. odora inflorescence, this diel vertical movement (“daytime descent to the female zone, nighttime ascent to the male zone”) is rhythmically repeated one to three times across the entire female stage. The female florets of A. odora are densely packed in a continuous cylindrical arrangement at the lowermost portion of the spadix. A single pollinator visit is unlikely to deposit pollen uniformly across all stigmas, inevitably creating pollination blind spots and uneven pollen distribution. Previous work has documented that natural fruit set in A. odora inflorescences is remarkably high, even significantly exceeding that obtained by artificial cross‐pollination; uneven manual pollen application is suspected to be the primary cause of this discrepancy (Ding et al., 2024). We therefore propose that the cyclical diel vertical movement of the two Colocasiomyia species during the female stage effectively compensates for this structural constraint, achieving uniform pollen deposition across the entire female zone through repeated inter‐zonal movements.

Subsequently, during the male stage of anthesis (which typically commences at night following the female stage), the flies move to the male zone in the upper part of the inflorescence, where pollen adheres to their body surfaces. The upper inflorescence maintains thermogenesis throughout the male stage, but the constriction of the spathe at the mid‐section prevents the flies from re‐entering the lower, cooler chamber. Given that Colocasiomyia flies cannot consume the pollen of A. odora, foraging requirements provide no incentive for them to remain in the upper inflorescence; thus, feeding behavior makes no contribution to pollen exportation. Instead, their behaviors during this stage are governed by the interplay between thermal repulsion and oviposition requirements. Because the two Colocasiomyia species differ in their oviposition requirements, they show divergent behaviors. C. xenalocasiae females deposit all their eggs in the lower part of the inflorescence and lack oviposition sites in the male zone; consequently, without a reproductive incentive to stay, these flies may be thermally repelled by the high daytime temperatures of the male zone and depart from the inflorescence entirely. Thus, C. xenalocasiae individuals, with their single upward movement, have only one chance to acquire pollen on their bodies. In contrast, C. alocasiae females preferentially oviposit in the male zone of the upper inflorescence. Thus, at least C. alocasiae females are less likely to be immediately driven away by the heat. To cope with the high daytime temperatures, they may instead relocate to the inner surface of the adjacent spathe as a thermal refuge. Similar to the female stage, the male stage of A. odora also lasts 1–3 d. This implies that C. alocasiae females may relocate repeatedly—driven by suitable thermal cues—between the male zone surface and the inner spathe surface one to three times. Previous studies on other populations in Japan and China have similarly reported that C. alocasiae tends to remain on male‐stage inflorescences, whereas C. xenalocasiae departs immediately (Toda et al., 2022). Evidently, this prolonged residence and repeated localized movement of C. alocasiae females enhance their potential to accumulate pollen on their bodies.

In summary, we conclude that the sophisticated pollination system of A. odora is driven by a synergistic interplay of floral thermogenesis, pollinator feeding, and oviposition requirements. During the female stage, while feeding requirements establish a strong preference for the sterile zone, it is the diel thermal gradient that acts as the primary engine for pollination. This thermal push–pull mechanism drives the vertical movement across the lower floral zones, ensuring uniform pollen importation. Superimposed on this thermally driven movement, the spatial alignment of feeding and oviposition sites in the lower chamber makes C. xenalocasiae the greater contributor to pollen importation. Conversely, during the male stage, feeding has no effect, and departure behaviors are determined by species‐specific oviposition preferences. The absolute requirement to oviposit in the male zone overrides thermal repulsion for C. alocasiae, leading to prolonged residence and repeated localized movements. Consequently, this specific oviposition preference renders C. alocasiae the dominant contributor to pollen exportation.

Floral thermogenesis‐driven pollination: Ecological and evolutionary perspectives

Our literature survey of published studies reporting measurable floral thermogenesis (temperature increase ≥ 0.5°C) identified 15 families (147 species across 52 genera) (Table S3; Thien et al., 2000; Salzman et al., 2020; Hernández‐Vera et al., 2021). Among the 147 species, 76 (51.7%) are dioecious, 44 (29.9%) are monoecious, and the remaining 27 (18.4%) are hermaphroditic. This compilation reveals a striking prevalence of spatiotemporal separation of sexual organs across thermogenic species (Hernández‐Vera et al., 2021; Peris et al., 2024): 90.5% of surveyed species (133/147) show spatial separation of male and female structures (either via dioecy or monoecy), and 70 of the 71 species with sequential sexual maturation are protogynous (the female stage precedes the male stage) (Núñez‐Avellaneda and Rojas‐Robles, 2008). This indicates that the vast majority of thermogenic species require pollinators to move between specific sexual structures at appropriate times for successful reproduction (Roemer et al., 2022).

Another important feature is the relationship between thermogenic plants and their pollinators. Among the 87 species for which pollinator information was available, 61 (70.1%) are pollinated by Coleoptera, 28 (32.2%) are pollinated by Diptera, and 4 (4.6%) are pollinated by Thysanoptera. More importantly, whether the relationship is mutualistic or deceptive, the pollinators consistently show prolonged residence within the inflorescence. In mutualistic, brood‐site pollination systems involving Coleoptera and Diptera, pollinators feed, mate, and oviposit inside the flower, often remaining for more than a day (Bröderbauer et al., 2014; Ibáñez et al., 2025). In deceptive systems, such as those in some Araceae and Hydnoraceae, trap‐type floral chambers confine pollinators for typically over a day until the male phase begins (Albre et al., 2003; Sayers et al., 2021). This prolonged stay, lasting hours to days, stands in stark contrast to the brief visits (seconds to minutes) typical of common hymenopteran or lepidopteran pollinators (Jones et al., 1998; Bauder et al., 2015). The extended residence of pollinators within the microclimates shaped by thermogenic flowers highlights the potential for movement between female and male zones and underscores the possible importance of thermogenesis in facilitating such within‐inflorescence pollinator movement.

Overall, we compare and summarize two patterns by which floral thermogenesis may shape microclimates to promote pollinator movement between sexual organs: movement between inflorescences and movement within an inflorescence. The first pattern—movement between inflorescences—is well documented in cycads (Zamiaceae), where thermogenic and volatile peaks facilitate pollinator transfer between separate male and female cones. For example, in Macrozamia lucida, a midday thermogenic peak in male cones, combined with high volatile concentrations, expels thrips, which then carry pollen to female cones (Terry et al., 2007, 2014). In Zamia furfuracea, beetles depart from male cones after an evening thermogenic peak and subsequent cooling, and are attracted approximately three hours later by heat radiation from thermogenic female cones (Salzman et al., 2020; Valencia‐Montoya et al., 2025). This thermogenesis‐mediated movement between conspecifics likely also operates in other dioecious or hermaphroditic thermogenic plants (Patino et al., 2000; Franz, 2007). The second pattern—movement within an inflorescence—is exemplified by Alocasia odora, where thermal gradients guide flies between female and male zones (Angioy et al., 2004; Ivancic et al., 2008). Beyond mutualistic aroids, similar within‐inflorescence dynamics may occur in deceptive Araceae (Bröderbauer et al., 2012; Sayers et al., 2021) and in plants with spatially separated bisexual flowers such as Hydnora and Prosopanche (Hydnoraceae) (Seymour et al., 2009; Rocamundi et al., 2024).

The recurrence of these two patterns across phylogenetically distant lineages raises the question of why temperature has been repeatedly co‐opted as a cue for directing pollinator movement. Generally, as ectothermic animals, insects critically rely on thermal sensation to drive behaviors that regulate their body temperature (Dillon et al., 2009; Gallio, 2026). Previous studies have shown that both innate heat avoidance and heat‐seeking behaviors are hard‐wired and robustly triggered by thermal cues (Simões et al., 2021; Capek et al., 2025), illustrating how microclimate temperature directly drives insect movement. Although these studies were conducted in Drosophila, the neural mechanisms of sensory perception and sensory‐guided behaviors are highly conserved among invertebrates (Carta et al., 2026). Thus, we suggest that the sensory‐guided movement of Colocasiomyia, a close phylogenetic relative of Drosophila, within the inflorescence of Alocasia odora, is also mediated through innate heat avoidance and heat preference behaviors. While floral color and scent cues can also mediate insect attraction or repulsion, insects can readily learn to associate these cues with reward or punishment, potentially overriding innate preferences through associative learning (Wright and Schiestl, 2009; Zhang and Luo, 2024). In contrast, it remains unclear whether associative learning can similarly override strict thermal preferences. Therefore, temperature serves as a reliable signal: The thermally guided behaviors of pollinators remain highly predictable, and this signal cannot be weakened or altered through acquired learning (Zars and Zars, 2006; Destierdt et al., 2026). By activating thermogenesis in specific floral regions at specific times, plants impose a dynamic thermal landscape that mediates pollinator movement within the inflorescence to align with the plant's sexual sequence. Furthermore, the broader comparative evidence—where thermogenesis is combined with the spatiotemporal separation of sexual organs and extended pollinator residence across at least 15 plant families—strongly suggests that microclimate‐guided pollinator movement represents a widespread, yet underappreciated pollination strategy.

CONCLUSIONS

Our study reveals that the microclimates generated by thermogenic flowers around their reproductive structures are crucial for pollination success. In Alocasia odora, the spatiotemporally dynamic microclimate functions akin to a “thermal map”, guiding pollinator movement within the inflorescence and thereby overcoming the challenge posed by its spatially and temporally separated sexual organs. Crucially, this thermal navigation, when coupled with the insects' feeding and oviposition requirements, dictates that the distinct movement trajectories of the two Colocasiomyia species contribute differentially to the reproductive success of A. odora. Specifically, the thermally constrained downward movement of C. xenalocasiae renders it the primary mediator of pollen import, whereas the localized upper‐chamber movement of C. alocasiae establishes it as the dominant agent of pollen export. Having qualitatively established the crucial role of this microclimate‐driven navigation, we explicitly note that future quantitative assessments of pollen transfer efficiency will further elucidate its contribution to reproductive fitness. However, the importance of such directed pollinator movement clearly extends beyond analogous aroid systems. The prevalent coupling of floral thermogenesis with the spatiotemporal separation of sexual organs, combined with the prolonged residence of pollinators, underscores a broad functional significance. We propose that microclimatic guidance represents a widespread, yet underappreciated, evolutionary strategy in which plants orchestrate complex insect behaviors to maximize reproductive success.

Limitations of the study

In the context of the multimodal nature of sensory ecology, our study identifies temperature as the dominant cue driving pollinators' diel vertical movement within Alocasia odora inflorescences, with two key limitations to note. First, while temperature is the primary driver in this system, it likely acts in concert with other synergistic or redundant cues: Floral scent may contribute to long‐distance pollinator attraction, and yet, our data provide no empirical support for it as the key signal guiding within‐inflorescence movement, and its potential functional role cannot be definitively ruled out. Additional untested factors include humidity gradients, and CO₂ concentration, which may also shape pollinator positioning and pollination precision. Second, we did not assess the influence of pollinators' endogenous circadian rhythms on movement patterns; diel differences in feeding, mating, and oviposition behaviors may interact with thermal and other environmental cues to jointly structure the spatial distribution of flies within the inflorescence.

MATERIALS AND METHODS

Study model and subject details

Study sites

This study was conducted in Haikou, Hainan Province, China (19°32′–20°05′ N, 110°10′–110°41′ E), characterized by a tropical monsoon climate with cool, humid winters and hot, rainy summers. We studied the population of Alocasia odora in Baishamen Park, which comprised over 1,000 individuals of A. odora. All experiments were carried out during peak flowering from November to December in 2021–2024.

Study species

Alocasia odora is a common understory monoecious perennial with protogynous spadices. Along the spadix, four distinct regions can be distinguished from top to bottom: the appendix (non‐reproductive), the male (staminate) zone, the sterile zone composed of sterile male flowers, and the female (pistillate) zone. The spadix is enclosed by a spathe and can be divided into an upper part (appendix and male flowers) and a lower part (sterile and female flowers) by a constricted area in the middle of the spathe. Anthesis begins with spathe opening and receptivity of female flowers (female stage), during which the upper part (appendix and male zone) initiates thermogenesis. After 1–3 d, at night, the constricted area of the spathe closes around the lower part, marking the end of the female stage and the onset of the male stage (Bian et al., 2021).

Three Colocasiomyia species are known to specialize on A. odora hosts. In the population studied here, only C. alocasiae and C. xenalocasiae were present, while a third species, C. grandis, has thus far been recorded only in Yunnan Province, China (Toda et al., 2022). The two pollinators occurring in our study system were identified based on established morphological characters, particularly costal bristle patterns (Takano et al., 2021).

Observation of adult feeding behavior of the two Colocasiomyia species

To investigate the feeding behavior and preference of Colocasiomyia alocasiae and C. xenalocasiae for different regions of Alocasia odora inflorescences, we conducted behavioral observation and distribution assays. First, adult flies collected from A. odora inflorescences were briefly immobilized via cold anesthesia on ice. We then identified the flies to the species level under a stereomicroscope (TS‐30H, PDV), and separated sexually mature males and females of each species based on morphological characters described in previous studies (Okada, 1990; Takenaka et al., 2006).

Alocasia odora spadices used in the experiments were at the female anthesis stage. To eliminate potential odor interference from the spathe and appendix on fly behavior, we removed these structures with a scalpel, retaining only the male zone, the sterile zone, and the female zone. To prevent the spadix from rolling during observation, we cut it longitudinally into two halves and placed the two halves into a single clean fly culture tube (24 mm diameter and 95 mm height). The tube was then positioned horizontally to facilitate observation of feeding behavior.

Each tube was inoculated with 30–60 flies of the same species and sex. We continuously recorded feeding, oviposition, and other behaviors of the flies via video recording. The experimental environment was maintained at 25°C with 60% relative humidity under constant artificial illumination. Based on preliminary observations, flies required approximately 1 h to recover from cold anesthesia and reach a stable behavioral state after introduction into the tube. Therefore, we counted the number of flies on each floral functional zone at 1 h post‐introduction. The proportion of flies in each zone was calculated as follows: (number of flies in a given zone)/(total number of flies across all floral functional zones). Individuals remaining on the tube wall or stopper were excluded from the counts. Separate experimental groups were set for males and females of each species, with 3–6 biological replicates per group.

Oviposition site preference of the two Colocasiomyia species

To determine whether there are interspecific differences in oviposition site selection between female C. alocasiae and C. xenalocasiae on different functional zones of the spadix, we conducted two independent oviposition preference experiments: One under controlled laboratory conditions and the other in the natural field habitat. The collection, species identification, and sex determination of female flies were performed as described above for the feeding behavior observations. For the inflorescences, we selected healthy A. odora inflorescences at the bud stage with unopened spathes. Each inflorescence was enclosed in a 100‐mesh (150 μm pore size) nylon net bag, which was tightly sealed to completely exclude external insects and prevent premature oviposition or other interference, until the female stage, when they were used as experimental materials.

Laboratory experiment: Intact A. odora spadices at the female stage were collected. After removing the spathe, each spadix was cut longitudinally into two halves and placed into a new fly tube (same specifications as above). Ten female flies of either C. alocasiae or C. xenalocasiae were introduced into each tube and allowed to oviposit freely on the spadix. The experiment was conducted under complete darkness at a constant temperature (25°C) and 60% relative humidity for 24 h. After the experiment, the spadix was removed and the number of eggs laid on each floral zone was counted under a stereomicroscope (TS‐30H, PDV). Eggs laid on the tube wall or stopper were excluded from the count. Independent experimental groups were set up for each species, with 10 biological replicates per group.

Field experiment: The field experiment was conducted in the natural population of A. odora. Intact inflorescences at the female stage were used. The net bag was carefully opened without damaging the inflorescence, and 10 female flies of the same species were quickly introduced into the bag, ensuring that the flies landed on the inflorescence. The bag was then immediately sealed to prevent fly escape and entry of foreign insects. After the flies had oviposited under natural conditions for 24 h, the inflorescence was collected and brought back to the laboratory. The number of eggs on each floral zone was counted under a stereomicroscope. Ten inflorescences (replicates) were used for each species.

Generation of floral RNA samples and data analysis

To identify genes involved in thermogenesis of Alocasia odora, we collected spadix samples from both the female zone and the male zone at two developmental stages: the bud stage and the female stage. Tissue samples (approximately 1 cm³ each) were obtained from three independent inflorescences per stage, resulting in a total of 12 samples—six from the bud stage and six from the female anthesis stage, with each inflorescence contributing both a female‐zone and a male‐zone sample, thus constituting three biological replicates per stage and zone combination. Following collection, RNA was extracted, purified, and used to construct sequencing libraries. Paired‑end RNA sequencing was performed on the Illumina platform (Illumina HiSeq™ 4000, CA, USA) at Personalbio Co., Ltd. (Shanghai, China).

De novo transcriptome assembly was performed using Trinity (v2.13.2 or later) with default parameters. Clean reads from all 12 samples were pooled for assembly in the absence of a reference genome. The resulting transcripts were clustered into unigenes using Trinity's built‐in clustering step. Functional annotation of unigenes was conducted by sequence similarity searches against the NCBI non‐redundant protein sequences (NR), Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), eggNOG (evolutionary genealogy of genes: Non‐supervised Orthologous Groups), Swiss‐Prot, and Pfam (Protein family) databases using BLASTX, BLAST2GO, and HMMER with E‐value ≤ 1e−5. Based on their established critical roles in plant thermogenesis, we specifically screened for genes encoding Alternative Oxidase (AOX) and Uncoupling Protein (UCP) (Costa et al., 2014; Yu et al., 2025).

Transcript abundance was estimated using RSEM software, and gene expression levels were normalized as transcripts per million (TPM) to allow for cross‐sample comparisons of relative abundance. For differential expression analysis, raw read counts were used as input for the DESeq. 2R package, which utilizes a negative binomial distribution model to estimate variance‐mean dependence. To identify genes significantly associated with thermogenesis, we performed pairwise comparisons between the thermogenic tissue (female stage‐male zone) and the non‐thermogenic control (bud stage‐male zone and female stage‐female zone). Differential expression results were reported as Log₂(Fold change) and Benjamini–Hochberg‐adjusted P‐values (padj). Genes were considered significantly differentially expressed at a threshold of adjusted P < 0.05 (Table S2).

Temperature measurement and fly movement

Based on preliminary observations suggesting that Colocasiomyia flies show temporal preferences for different parts of the Alocasia odora inflorescence throughout the day, we investigated the potential correlation between fly movement patterns and temperature variation on the spadix.

To this end, we conducted simultaneous observations of fly activity on three synchronously flowering inflorescences, starting at the female stage and spanning approximately 3 consecutive days. A single four‐channel temperature logger (DT‐3891G, CEM), equipped with four K‐type thermocouple probes (WRNK‐191), was used; however, only one focal inflorescence could be continuously monitored for temperature. On this inflorescence, temperatures of the appendix, the male zone, and the female zone were recorded every 3 min by inserting each probe about 1 cm into the corresponding tissue; the fourth channel simultaneously recorded ambient air temperature as a control. To confirm the consistency of thermogenic patterns, we also measured temperatures on three additional inflorescences (different individuals) on separate dates using the same method.

Concurrently, fly activity was observed and counted on all three inflorescences. At hourly intervals, with minimal disturbance to the flies, we recorded the number of individuals present in the upper parts (male + appendix + upper spathe) and lower parts (sterile + female + lower spathe) of each spadix, allowing calculation of the total number of flies and their distribution ratio per time point. To enable consistent observation of the lower inflorescence when it was naturally enclosed by the spathe, we slightly widened the constricted area of the spathe during the female stage. This modification allowed flies to move into the lower part during the male stage as well; however, it did not noticeably alter their foraging sequence, as flies consistently landed first on the upper part before moving downward. We acknowledge that this manipulation inevitably altered the natural humidity and gas concentration gradients in the lower part of the inflorescence. However, this modification was necessary to observe within‐inflorescence movement during both the female and male stages. Critically, it was precisely under these altered conditions—where humidity and volatile gradients were presumably reduced—that we observed the regular activity of Colocasiomyia flies and their strong responses to temperature manipulations, providing strong evidence for the dominance of thermal cues.

Field‐based temperature and light manipulation assays

To test the causal effects of temperature and light on fly movement, we conducted field manipulation experiments.

The overall aim of the temperature adjustments was to shift tissue temperature toward or away from the 20°C–25°C range. For temperature manipulation, we conducted separate experiments during the day (12:00) and at night (24:00, n = 10 inflorescences per period). During the experiments, we first recorded the baseline fly distribution on the inflorescence and the temperatures of the female and male zones using the aforementioned four‐channel temperature detector, with two probes inserted into the surface tissues of the female and male zones, respectively. We selected 12:00 and 24:00 for manipulation experiments because these time points represent the thermal extremes (daytime maximum and nighttime minimum) and the stable endpoints of fly distribution, minimizing confounding effects from ongoing transitional movement.

We then sequentially applied two manipulations to the same inflorescence. During the day, the sequence was as follows: (i) cooling the upper part (to 22°C ± 2°C) and (ii) heating the lower part (to 30°C ± 2°C). At night, the procedure was similar, but with the sequence reversed as follows: (i) heating the lower part (to 25°C ± 2°C) and (ii) cooling the upper part (to 17°C ± 2°C). Heating was applied externally using USB heating pads powered by a power bank, and cooling was achieved with bagged ice. The devices were placed against the outer surface of the target spathe region without direct contact with the flies, thus avoiding direct disturbance that might prompt them to leave the inflorescence. In each case, the temperature was allowed to change gradually over approximately 5 min until the target temperature was reached, was held stable for 30 min to allow sufficient time for flies to respond and move, and was then recorded along with the fly distribution before proceeding to the next phase. Temperatures and fly distributions were recorded in each phase. The maximum temperature that we applied to the inflorescence was approximately 30°C, which falls within the highest temperature range naturally achieved by the appendix and male zone of Alocasia odora during winter in our study population (Bian et al., 2021).

For light manipulation, sample sizes were n = 13 for daytime shading and n = 6 for nighttime illumination. Inflorescences were covered with a black box during the day (12:00–14:00) to simulate darkness and illuminated from above with daylight lamps (2,000 Lux; GMETAL BD1600L) at night (00:00–02:00). Each manipulation was maintained for 30 min, and fly distribution was recorded before and after the treatment.

Floral scent collection

To investigate the diel variation in volatile organic compound (VOC) emission dynamics of Alocasia odora inflorescences and to provide samples for subsequent electrophysiological experiments, we collected volatiles from inflorescences and ambient air (control) using dynamic headspace adsorption with a pull system. Six inflorescences at the female stage were consecutively sampled at four time points in situ: 12:00 on Day 1 (female stage), 24:00 on Day 1, 12:00 on Day 2, and 12:00 on Day 3 (male stage). Each entire inflorescence or an equivalent volume of control air was enclosed in a polyester oven bag (25 × 38 cm, Toppits®). Purified air, obtained by passing ambient air through a charcoal filter cartridge (50/100 mg filling, Beijing Municipal Institute of Labour Protection), was drawn through the system at 100 mL/min via a vacuum pump (QC‐1B, Beijing Municipal Institute of Labour Protection). Volatiles were trapped onto an adsorbent tube packed with a thin layer of activated carbon (CLSA, 1.5 mg, Brechbühler AG) for 3 h (Brodmann et al., 2008). Adsorbed volatiles were eluted with 50 μL of dichloromethane (Sigma‐Aldrich, GC grade) and stored at −20°C. After each sampling session, the adsorbent tubes were cleaned three times with ethanol, hexane, and dichloromethane (Sigma‐Aldrich, GC grade), and then dried and reused for subsequent sampling.

To investigate the distribution of volatile compounds in different regions of the inflorescence, we sampled dissected floral tissues using solid‐phase microextraction (SPME). Two first‐day female‐stage inflorescences from different individuals were used, along with two ambient air samples (collected five meters away from the inflorescences) that served as controls. After complete removal of the spathe, each inflorescence was manually dissected into four parts: the appendix, the male zone, the sterile zone, and the female zone. Each part was separately placed in a polyester oven bag (as above). After a 10‐min equilibrium period, an SPME fiber (50/30 μm DVB/CAR/PDMS, 1 cm fiber length, 57328‐U, Supelco) was inserted into the headspace to adsorb volatiles for 30 min. The sampled fibers were immediately transported to the laboratory and manually introduced into the GC injector port, which was set at 280°C, for desorption over 15 s in splitless mode.

Chemical analysis

Floral scent extracts were analyzed using a gas chromatography system coupled with a mass spectrometer (GC–MS, Agilent 7890 A), equipped with a capillary column (30 m × 0.25 mm × 0.25 μm, HP‐5MS UI, Agilent). Helium was used as the carrier gas at a constant flow rate of 1 mL/min. A 2 μL sample was injected in on‐column mode at an injector temperature of 280°C. The oven temperature program was as follows: Initial temperature held at 40°C for 2 min, then increased at 12°C min⁻¹ to 300°C, and finally held at 300°C for 3 min. The MS source temperature was maintained at 230°C, and ionization was performed in electron impact (EI) mode at 70 eV.

Compounds were tentatively identified by comparing their mass spectra with the NIST 2011b mass spectral library and further confirmed by matching their experimental Kovats retention indices (C5–C25 n‐alkanes, Sigma‐Aldrich) with reference values in the NIST database (van Den Dool and Dec Kratz, 1963). Five compounds released by the inflorescence—methyl propionate, methyl butyrate, methyl 2‐methylbutyrate, methyl benzoate, and methyl salicylate (obtained from commercial sources at the highest available purity)—were additionally verified by direct comparison with synthetic standards (based on both mass spectra and retention times; see Table S1).

Gas chromatography‐electroantennographic detection (GC‐EAD)

To preliminarily screen for volatile components that are specifically detected by the antennae of Colocasiomyia flies, we performed GC‐EAD analysis. The GC conditions were essentially identical to those described for the GC‐MS analysis above. The column effluent was split 1:1 between the flame ionization detector (FID) and a heated, humidified nitrogen gas stream directed to the Syntech EAD system (Agilent 7890A‐5975C/EAD). A detached female fly head was used for recording; one Ag–AgCl glass electrode filled with saline was inserted into the posterior part of the head, while another electrode was connected to one randomly selected intact antenna. The signals were amplified and transmitted to a signal acquisition interface (IDAC‐2, Syntech). Signals from the FID and the antenna were recorded simultaneously using GC‐EAD software (v4.6.1, Syntech). Volatile collections from the appendix parts (collected separately using the method described above) were collected from three female individuals each of C. alocasiae and C. xenalocasiae.

Field four‐choice assays

To test the long‐range attractiveness of volatiles from different regions of the Alocasia odora inflorescence to Colocasiomyia flies, we conducted a four‐choice assay outdoors. Four black plastic cups (10 cm diameter × 10 cm height) were placed in a 2 × 2 arrangement on the ground near a natural Alocasia population, with a spacing of 15 cm between adjacent cups. Each cup contained one of the following four tissue types dissected from the same female‐stage inflorescence: the appendix, the male zone, the sterile zone, or the female zone (spathes were completely removed). An entrance hole 1 cm in diameter was made in each cup. After 24 h, the Colocasiomyia flies attracted to each trap were collected and counted (30 independent trials; total captures = 155 flies). Additionally, our behavioural assays were conducted under ambient field conditions where temperature was not controlled; therefore, we cannot exclude the possibility that temperature differences among floral zones influenced volatile emission rates and thus indirectly affected fly attraction in these assays.

Laboratory two‐choice assays

To further investigate the close‐range odor preference of Colocasiomyia flies for different regions of the inflorescence, we performed two‐choice assays. A transparent acrylic flight chamber (20 cm long × 10 cm wide × 10 cm high) was used. Two 5 mL centrifuge tubes, placed approximately 10 cm apart, served as traps inside the chamber. Trap 1 contained no plant tissue (empty control). Trap 2 contained a tissue block (approximately 3 cm³) from a female stage inflorescence, excised using a scalpel. The lid of each tube had a hole (approx. 0.5 cm diameter) allowing fly entry. Four experimental groups were set up, corresponding to the four inflorescence regions: the appendix, the male zone, the sterile zone, and the female zone (n = 15 replicates per group). Flies were randomly collected from inflorescences, and 30 flies were introduced into the flight chamber using an aspirator. Flies were allowed to move freely within the chamber for 24 hours under dark conditions at room temperature. After the experiment, the number of flies in Trap 1 and Trap 2 was counted separately. The Preference Index (PI) was calculated as follows: PI = [(Number of flies in Trap 2 − Number of flies in Trap 1)/(Number of flies in Trap 1 + Trap 2)] × 100% (Wang et al., 2014).

Comparative analyses of thermogenic flower traits

To explore the potential role of floral thermogenesis in creating microclimatic heterogeneity and influencing pollinator movement, we conducted a systematic literature search on November 22, 2025, using the Web of Science database with the query “thermogenesis AND (pollination OR pollinator)”. The search yielded 278 records. We retained only studies that reported actual temperature measurements of individual flowers or inflorescence regions (with a temperature increase ≥ 0.5°C) (Ito‐Inaba et al., 2019) and excluded non‐floral thermogenesis unrelated to pollination (e.g., Santalaceae) (Ross Friedman et al., 2013). After manual screening, 105 relevant articles were included for analysis.

From each article, we extracted the following information: family and species name, sex system (spatial separation of sexes: Plants with unisexual/bisexual flowers, and dioecious plants), dichogamy (temporal separation of sexes: protogyny or protandry), thermogenic position (male/female zone), thermogenic timing (phases or peaks), pollinator identity (specific species or genus) and type (higher taxonomic group, e.g., Diptera, Coleoptera), interaction type (mutualism/deception), whether the effect of plant thermogenesis on pollinators was experimentally studied (yes/no), putative thermogenic function proposed by the authors (BSR: boost floral scent release; DA: direct attraction; HR: heat reward; EPF: ensure the plant reproductive function; and PPM: promote the pollinator movement), and our putative pollinator movement pattern (within or between inflorescences).

Statistical analyses

Statistical analyses and graph plotting were performed using GraphPad Prism (version 9.5.1) and Microsoft Excel 365 (excluding transcriptome data), while image optimization of the generated figures was conducted with Adobe Photoshop (version 26.4.1). Parametric tests were applied to normally distributed data sets: Two‐tailed t‐tests assessed pairwise differences in field light experiments, while one‐way ANOVA with Tukey's post hoc tests analyzed multi‐group comparisons in field temperature manipulation experiments. Non‐parametric alternatives were used for non‐normally distributed data: Preference indices from laboratory two‑choice assays were analyzed using Wilcoxon signed‑rank tests comparing the median PI against zero, treating each replicate as the experimental unit; Friedman tests with Dunn's corrections handled related multi‐group designs in field volatiles four‐choice assays; and Kruskal–Wallis tests with Dunn's comparisons addressed independent multi‐group analyses in diel variation in floral scent emission studies.

Generative AI usage statement

During the preparation of this work, the authors utilized DeepSeek and Gemini to assist in improving language clarity and fluency. The authors carefully reviewed, edited, and validated all AI‐generated suggestions. The authors take full responsibility for the final academic content, interpretations, and conclusions presented in this work.

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Y.J.Y., W.F.Z. (partly), and Y.B.L. conceptualized the study. Y.J.Y., X.D., and X.Q.S. performed field experiments. Y.J.Y. and X.D. analyzed data. Y.J.Y. and Y.L. performed RNA sequencing and analysis. Y.J.Y., W.F.Z., Y.L., and Y.B.L. wrote the manuscript. Y.B.L. and Y.L. acquired funding. All authors have read and approved the content.

Supporting information

Additional Supporting Information may be found online in the supporting information tab for this article: http://onlinelibrary.wiley.com/doi/10.1111/jipb.70318/suppinfo

Figure S1. Expression patterns of alternative oxidase (AOX) and uncoupling protein (UCP) genes across spadix developmental zones and stages

Figure S2. Antennal response of Colocasiomyia alocasiae to appendix volatiles

Figure S3. Temporal variation in volatile compound emissions from inflorescences

Table S1. Floral volatiles collected using SPME methods

Table S2. Expression of alternative oxidase (AOX) and uncoupling protein (UCP) genes across spadix developmental zones and stages

Table S3. Comparative analysis of thermogenic traits across lineages

Biographies

Yu, Y. , Luo, Y. , Zhang W., Ding, X. , Song, X. , and Luo, Y. (2026). Thermogenesis‐derived spatiotemporal microclimates guide pollinator movement to ensure pollination. J. Integr. Plant Biol. 68: 4071–4087.