2.1 Systematic Review

To build the database on pollinators of Aristolochia species and populations, we searched articles in Google Scholar (scholar.google.com), with the last search on 29 October 2024. We used the search strings: ‘Aristolochia’ AND ‘Pollination’. Then, we followed the PRISMA guidelines to screen the references (Preferred Reporting Items for Systematic Reviews & Meta-Analyses in Ecology and Evolutionary Biology, O'Dea et al. 2021).

The search returned 2850 results, which were screened by a single person. The order of the eligibility criteria was: (1) the title, abstract, or keywords included the search terms; (2) the study provided clear information on the collection and identification of dipteran pollinators of any Aristolochia population; and (3) the study reported the taxonomy of dipteran pollinators (e.g., order, family, genus, species).

For each Aristolochia population, we extracted the identification of the dipteran pollinators found inside the trap flowers considering the most refined taxonomic level, that is, up to order, family, genus, or species. We only included dipterans in our review since they are considered the exclusive pollinators of Aristolochia species so far (Berjano et al. 2009). We also extracted the geographic coordinates of the studied population and used the coordinates of the city's centroid through Google Earth (earth.google.com) when the authors did not provide the exact location. To reduce bias and simplify the many potential taxonomic levels, we considered the family for morphospecies identified up to subfamily and the genus for morphospecies identified up to subgenus or species group. The validity and spelling of scientific names of the final list were checked using the scientific names matching tool of the GBIF organisation (Global Biodiversity Information Facility; gbif.org/tools/species-lookup).

2.2 Data Processing

To demonstrate the lack of taxonomic refinement, we quantified the number of dipteran records identified only at the order, family, genus, and species levels. Additionally, we created a metanetwork using the R-software package bipartite version 2.19 (Dormann et al. 2008) to visually represent the interactions between dipterans identified at the distinct levels of refinement and the Aristolochia populations. To identify general patterns, we used presence-absence data (i.e., binary) because abundances were not available for all records, and studies have different sampling efforts. To demonstrate the taxonomic impediment in relation to the most refined taxonomic potential (i.e., species level), we calculated the percentage of identifications at the species level (Number of identified species/Total number of Diptera taxa identified × 100).

2.3 Study Case

To highlight the importance of taxonomists in identifying and unrevealing Diptera groups as pollinators, we used the published data from a Brazilian population of the South American Aristolochia esperanzae (Figure 3a; see Matallana-Puerto, Duarte, et al. 2024). We selected this species because it is one of those with the highest number of morphospecies recorded in Aristolochia so far. The dipterans were collected at the end of the first day of anthesis from 81 flowers when these were in the female phase. The dipterans were identified as morphospecies in a less in-depth screening work by a few Diptera taxonomists (see Matallana-Puerto, Duarte, et al. 2024). For the current study, the same Diptera individuals were identified by more taxonomists, covering a wider range of specialists in each Diptera subgroup. Then, the results of taxonomic refinement were compared in these two conditions: without and with the presence of taxonomists specialised in each Diptera family.

4.1 Overview

By synthesising information on the taxonomic refinement of dipteran pollinators of Aristolochia populations, we demonstrated that the level of taxonomic refinement is low since most records were identified to family and genus levels. Alternatively, there were few records identified at the species level, which would be the ideal conditions for a complete understanding of the biodiversity of Diptera pollinators. As a solution to this problem, we showed that collaborating with taxonomists led to increased taxonomic refinement, with more distinguished morphospecies and more precise identifications at the genus and species levels. This underscores the importance of interdisciplinary research in advancing our understanding of lesser-known dipteran pollinator groups. We discuss our findings below.

4.2 Why So Much Neglect?

Despite the wide range of lifestyles and ecological roles exhibited by dipterans, such as decomposers, phytophagous, parasitoids, predators, and pollinators (Skevington and Dang 2002; Wiegmann and Yeates 2017), studies suggest that their neglect in various ecological roles is common due to the high diversity of the order, as well as the limited research on their taxonomy and natural history (Skevington and Dang 2002; Ssymank et al. 2008; Mlynarek 2022). For instance, in pollination studies, although some Diptera families, such as Bombyliidae, Calliphoridae, Drosophilidae, Muscidae, Nemestrinidae, Tabanidae, Tachinidae, and Syrphidae, are popularly known to act as pollinators, this order remains less studied than more charismatic taxa (Symank et al. 2008; Inouye et al. 2015; Orford et al. 2015; Ollerton 2017). Even in recent years, lesser-known Diptera families, such as Anthomyiidae, Mycetophilidae, Sciaridae, Tephritidae, and Chloropidae, have been reported as potential pollinators, highlighting the limited taxonomic knowledge of Diptera pollinators and how much we have to discover (Raguso 2020). In this context, while the taxonomic impediment of the order and the limited ecological research may explain the limited taxonomic knowledge of dipteran pollinators (Orford et al. 2015), this may also be related to the lack of collaboration with taxonomists. Additionally, the ongoing global taxonomist crisis, characterised by a decline in the training of new professionals and the evasion of the existing ones, may further exacerbate this issue (Orford et al. 2015; Löbl et al. 2023).

Our results corroborate the taxonomic impediment associated with dipteran pollinators since few records of these in Aristolochia spp. flowers were identified at the species level (21.3%), while most were identified at the family or genus levels (41% and 35.8%, respectively). Furthermore, the mean percentage of species identification across Aristolochia populations was 34.11%, with most populations having < 5% of this maximum identification. Although it is known that other dipteran-pollinated plants, whether with similar or distinct pollination systems, also exhibit limited taxonomic refinement, with many flies identified only at the family or genus level (Freitas and Sazima 2006; Berjano et al. 2009; Bashir et al. 2013; Ollerton et al. 2017; Cook et al. 2020; Ren et al. 2023). Since our results refer to the pollinators of a single genus, caution should be taken when extrapolating our findings to avoid excessive generalisation.

4.3 Dipterans and Aristolochia

Based on the data obtained through the review of dipterans associated with Aristolochia, this section summarises our findings and discusses their biological significance, highlighting how taxonomic refinement of pollinators can be used to understand pollination mechanisms based on the natural history of specific Diptera groups. We found that the trap flowers of 40 populations of 23 Aristolochia species interact with 61 species, 109 genera, and 43 families of Diptera. Comparatively, the trap flowers of 69 taxa of Ceropegia interacted with only 25 Diptera families (Ollerton et al. 2017), highlighting the high diversity of potential pollinating Diptera species that may be hidden within the trap flowers of Aristolochia. The Diptera families interacting with Aristolochia represent 35% of the 180 Diptera families (Bertone and Wiegmann 2009; Mlynarek 2022). Some Aristolochia species show specialised interactions with a single pollinator species (e.g., A. inflata; Sakai 2002) or with multiple species from the same genus or family (e.g., A. littoralis, A. bianorii; Hall and Brown 1993; Alpuente et al. 2023). On the other hand, other species interact with multiple taxa involving different Diptera families (e.g., A. baetica, A. manshuriensis, A. esperanzae; Nakonechnaya et al. 2021; Matallana-Puerto, Duarte, et al. 2024; Rupp et al. 2024). This specialisation-generalisation range suggests that Aristolochia species vary in deceptive strategies employed to attract their pollinators, as well as their ecological conditions.

On the other hand, the high and taxonomically underestimated diversity of Diptera underscores the relationship between different pollinator taxa and Aristolochia, which seemingly exploit diverse ecological behaviours of dipterans through distinct deceptive strategies involving scents and/or colours (Oelschlägel et al. 2015; Martin et al. 2017; Rupp et al. 2021, 2024; Alpuente et al. 2023). In agreement, most Diptera families interacting with Aristolochia exhibit diverse habits (Skevington and Dang 2002). For instance, Culicidae and Drosophilidae are known as nectar feeders, Empididae and Ceratopogonidae can be pollen feeders, while Syrphidae can consume both resources (Skevington and Dang 2002). However, Aristolochia trap flowers seemingly do not indicate the presence of such resources, suggesting that representatives of these families are deceived through mimicry mechanisms (Johnson and Schiestl 2016; Lunau and Wester 2017). In fact, most Diptera families trapped within Aristolochia spp. are associated with decaying organic matter, either for adults to feed on or as an oviposition site (Skevington and Dang 2002). This may include decaying plant material like wood, flowers, and fruits; decaying animal material like faeces and carrion; and fungi (Skevington and Dang 2002). Some of these families can also have species acting as parasitoids and/or predators of larvae and small invertebrate insects (Skevington and Dang 2002). Interestingly, the families that interacted with the greatest number of Aristolochia species and populations (Phoridae, Drosophilidae, and Sciaridae) are also associated with a wide diversity of decaying organic matter (Disney 2012; Skevington and Dang 2002). Even, some Phoridae and Drosophilidae species can also be parasitoids or predators of other insects (Skevington and Dang 2002). This suggests a high diversity of floral traits of Aristolochia species that can exploit the behaviour of several Diptera lineages. Specifically, this may be related to the high diversity of volatile compounds emitted by Aristolochia trap flowers, which may mimic yeast-fermented fruits, vertebrate and invertebrate carrion, animal faeces, or even insect semiochemicals that can attract kleptoparasitic flies (Johnson and Jürgens 2010; Oelschlägel et al. 2015; Martin et al. 2017; Rupp et al. 2021, 2024; Alpuente et al. 2023).

Diptera-mediated deceptive pollination systems are also prevalent in other unrelated trap flowers, such as Arum (Araceae), Arisaema (Araceae), and Ceropegia (Apocynaceae) (Gibernau et al. 2004; Vogel and Martens 2000; Ollerton et al. 2017). Although dipterans are obligate pollinators in these specialised pollination systems, recent studies with these plants continue to discover pollinating flies belonging to rarely known groups as pollinators (Raguso et al. 2020). Such is the case of Chloropidae, known for containing some kleptoparasite groups (e.g., Trachysiphonella, Conioscinella, Tricimba), which were recently recognised as pollinators of trap flowers belonging to the unrelated genera Aristolochia and Ceropegia (Heiduk et al. 2015; Oelschlägel et al. 2015). Interestingly, the trap flowers of these unrelated genera produce volatile compounds associated with insects being preyed upon. This taxonomic recognition, coupled with ecological data, enabled experiments that led to the discovery of a new pollination system known as kleptomyophily (Heiduk et al. 2015; Oelschlägel et al. 2015). This highlights how much we still must discover about lesser-known dipteran pollinator groups and the importance of their refined identification, even in understanding fascinating cases of evolutionary convergence.

4.4 What Can We Do?

In our case study with A. esperanzae, we demonstrated that collaborating with taxonomists specialised in different Diptera subgroups can enhance the refinement of pollinator identification, uncovering new biological information, as well as new records that will potentially reveal new species (Costa et al., in prep.). In this case, the number of morphospecies increased by 53.75% (from 37 to 80) and identification at the family level by 33.3% (from 14 to 21). Furthermore, we achieved the identification of 42 genera and 31 species that had not been previously identified. The insights gained through the taxonomic work with specialist taxonomists place A. esperanzae as the population interacting with the highest number of Diptera morphospecies and species worldwide to date (80 morphospecies and 31 species), followed by A. manshuriensis (36 and 16; Nakonechnaya et al. 2021) and A. baetica L. (39 and 15; Rupp et al. 2024). Our results reveal the hidden biodiversity within the community of Diptera pollinators and demonstrate how collaboration with taxonomists enhances taxonomic refinement, as the final number of identified morphospecies and species was higher when taxonomists were involved.

Furthermore, taxonomic refinement allowed us to identify the primary fly pollinators of the Brazilian population of A. esperanzae, which belong to Ulidiidae (Matallana-Puerto, Duarte, et al. 2024). This family is widely known for being a maize crops pest, developing in decomposing plant organic matter and animal faeces (Cruz et al. 2011; Goyal et al. 2012; Ebejer and Nicolosi 2022), but rarely recognised as pollinators. To date, only a few studies have reported some Ulidiidae morphospecies as potential pollinators of fetid flowers. Examples include species of the genus Acrosticta, recorded as potential pollinators of Acianthera aphthosa (Orchidaceae; Pansarin et al. 2016) and A. esperanzae in Argentina (Aristolochiaceae; Aliscioni et al. 2017), a species of Idana reported in another population of A. aphthosa in Brazil (Ribeiro et al. 2006), and a species of Timia considered the pollinator of Pilostyles haussknechtii in Iran (Apodanthaceae; Bellot and Renner 2013). In our study, which also involved fetid flowers from the A. esperanzae population in Brazil, the Ulidiidae were among the main pollinators (Matallana-Puerto, Duarte, et al. 2024). We recorded two morphospecies of Acrosticta and identified species of Euxesta (5 spp.), Notogramma (1 spp.), and Physiphora (1 spp.) as pollinators (Table S2). These findings underscore the importance of taxonomists in revealing the identity of lesser-known pollinators, such as Ulidiidae. Moreover, our findings suggest that pollination by species of this family may be more frequent than previously thought, particularly in flowers emitting volatile compounds associated with decomposing organic matter or animal faeces, such as in the understudied tropical Aristolochia species.

Notably, we were far from achieving 100% of morphospecies identification, even after collaborating with specialist taxonomists. This can be related to various general and group-specific taxonomic challenges. The first issue is that most specimens trapped by these flowers belong to dipteran families considered ‘open-ended taxa’, meaning they are highly speciose and widely distributed while their true diversity remains largely unknown (sensu Bickel 2010). This can be the case of some families commonly found in A. esperanzae, such as Tachinidae (1477 genera and 8592 species), Cecidomyiidae (736 and 6203), Syrphidae (209 and 6107), Muscidae (187 and 5200), Phoridae (300 and 4500), and Drosophilidae (74 and 4000) (Bickel 2010; Courtney et al. 2017). Furthermore, the lack of funding for studying certain animal groups and the reduced number of employed taxonomists hinder species identification (Marques and Lamas 2006; Haseyama et al. 2024). Consequently, the availability of taxonomic keys for dipteran families and genera is limited. Even in less diverse groups that were trapped by A. esperanzae trap flowers, such as Piophilidae (14 genera and 83 species) and Richardidae (34 and 74), there is a lack of taxonomic information sufficient to identify their species (Haseyama et al. 2024). A third issue is that species-level identification often requires males (Sinclair et al. 2013), whereas A. esperanzae captured predominantly females (Matallana-Puerto, Duarte, et al. 2024), further complicating species identification. Some genetic techniques, such as DNA barcoding, are effective in delimiting species and have gained increased use in recent years (Kranzfelder et al. 2017; Meier et al. 2024; Riccardi and Hartop 2024). However, the absence of reference libraries for diverse insect groups such as Diptera can lead to biased data, particularly in less funded tropical regions where a considerable portion of the biodiversity remains unknown (Meier et al. 2006, 2022, 2024; Virgilio et al. 2012; Kranzfelder et al. 2017; Kjærandsen 2022; Riccardi and Hartop 2024). Therefore, as highlighted by other studies, we believe that the first step to improving the recognition of the role of Diptera in ecological processes, including pollination, is to increase interdisciplinary collaboration with taxonomists (Kim and Byrne 2006; Granjou et al. 2014; Orford et al. 2015). In the future, with more specialist taxonomists, greater investment, and the broader application of genetic tools, we hope to improve the species identification of dipteran pollinators of A. esperanzae, as well as those of other plant species.