Wild bees are essential pollinators of both native and cultivated plants, but their populations are declining worldwide. Conservation efforts are hindered by insufficient data, especially in the Mediterranean basin, which hosts some of the most diverse pollinator communities in the world. Particularly in Sardinia, the second largest island in the Mediterranean, information on the bee fauna is still limited. The aim of this work was to provide the first checklist of Apoidea Anthophila from an unexplored peninsula in north-eastern Sardinia (Italy), by combining traditional (morphologically-based) taxonomy and DNA barcoding. In addition, records of flower visits are provided and shown in a visitor network to enrich the scarce data on the associations between wild bees and plants in the Mediterranean Region. Bees were sampled from April to October 2022–2023 with two Malaise traps and nets. DNA was extracted to amplify sequences of the mitochondrial gene Cyotochrome oxydase I, which were then compared with those in BOLD using the identification tool and by constructing neighbor-joining phylogenetic trees. Seventy-six different species belonging to 29 genera and six families were collected and identified. A total of 212 COI sequences were obtained for 61 different species, many of which had not yet been sequenced from Italian populations. Five of the collected taxa are Sardo-Corsican endemics, whereas six species are newly recorded from Sardinia. Finally, we highlight potential taxonomic issues and new flower visit records, emphasizing the need for further research to better understand the taxonomy and ecology of this diverse group of insects toward their conservation.

Checklist, COI, DNA barcoding, endemism, flower visitation networks, morphological identification, new records

Bees pollinate most of the wild and cultivated plant species, contributing to the maintenance of plant communities and agricultural production (Quaranta et al. 2004; Klein et al. 2007; Stine et al. 2015; Matias et al. 2017). The conservation of wild bees is critical for both natural ecosystems and dependent crops (Julier and Roulston 2009; Stine et al. 2015; Vaughan et al. 2015), yet several reviews have highlighted that populations are declining worldwide (Bartomeus et al. 2013; Burkle et al. 2013; Garibaldi et al. 2013). According to the IUCN European Red List of Bees, 9.2% of the assessed bee species are threatened with extinction and 5.2% are considered Near Threatened, but the status of most of the species (56.7%) remains unknown due to insufficient information (Nieto et al. 2014). Similarly, Quaranta et al. (2018) highlighted the lack of sufficient data for most Italian bee species, compromising a reliable assessment of their conservation status. Moreover, the associations between wild bees and flowering plants are insufficiently documented for many bee species, particularly in Mediterranean regions. The lack of information on floral associations, as with other ecological aspects, makes it difficult to develop effective conservation strategies.

The Mediterranean basin hosts some of the most diverse pollinator communities in the World and hosts highly diverse bee fauna (Michener 1979; Michener 2000; Nieto et al. 2014; Orr et al. 2020). Currently, published data on wild bees underline a significant gap in research efforts between the central-northern part of the European continent, where most of the taxonomic specialists are concentrated, and Mediterranean countries, which are characterised by greater species richness (Quaranta et al. 2018). This gap highlights the need to strengthen local expertise and invest in basic research on the biology and ecology of wild bees, which is essential for their effective conservation. Particularly in Sardinia, the second largest island in the Mediterranean, our knowledge of the bee fauna is still limited. This island is home to a unique flora and fauna, with many distinctive species, including endemic taxa shared with Corsica or other regions, as well as several species with a very restricted range. The checklist of Italian wild bees (Hymenoptera: Apoidea: Anthophila) of Comba (2019) reports 316 bee species in Sardinia, including several endemic taxa, such as Sardo-Corsican species or subspecies (e.g., Panurgus corsicus Warncke, 1972 or Bombus terrestris sassaricus Tournier, 1890), Tyrrhenian subspecies (e.g., Andrena rotundata diomedia Warncke, 1975 or Osmia ferruginea igneopurpurea Costa, 1882) and Italian subspecies (e.g., Melecta leucorhyncha taormina Strand, 1919). Recently, different new records and species have been reported for this island, such as Hoplitis occidentalis Müller, 2012, Tetralonia gennargentui (Nobile, Catania & Bella, 2021) and Andrena antonellae Praz & Genoud, 2022 (Müller 2012; Catania et al. 2021; Nobile et al. 2021; Praz et al. 2022). Moreover, certain areas of Sardinia such as the Culuccia Peninsula (see Mariottini et al. 2024) where the present study was carried out, lack information about the bee fauna. This small area of approximately 3 km² is located north-east of Sardinia (Italy), near the Maddalena archipelago, between Santa Teresa di Gallura and Palau (Fig. 1). This area shows a complex environmental mosaic that has remained relatively untouched by human activity for the past century, because of its minimal human presence compared to the neighbouring touristic coastal regions.

Figure 1. 

Position of the study area in Sardinia and in Italy (A). Satellite image of the Culuccia Peninsula and location of the sampling site (B), base map: orthophoto © 2022 Google. The red diamonds represent the position of the Malaise traps while the triangles are the main site (transects) sampled with nets. The different colours of the triangles indicate the Corine Land Cover classes in which the sampling sites were located. PAS=Pasture, MMA=Mediterranean maquis, TWS=Transitional woodland/shrub, HDV=Herbaceous dune vegetation, VIN=Vineyards.

The main aim of this study is to provide the first checklist of bees of the Culuccia Peninsula, considering the current limited knowledge on the inventory and distribution of wild bees in the Mediterranean region. In addition, the use of a modern integrative taxonomic approach (morphological and molecular) for the accurate identification of the samples allowed to obtain the first COI sequences of the Sardinian populations for most of the species studied. Finally, we provide records of bee visits to flowers to show the interaction patterns between bees and plants on the peninsula and to enrich the limited knowledge of plant-bee associations in northern Sardinia.

A total of 469 individuals were collected, 383 with nets and 86 with Malaise traps, belonging to 76 different species from 29 genera and six families (Table 1, Suppl. material 5: table S3). Of the total species, 43.4% were sampled exclusively with nets, 42.1% with both nets and Malaise traps, and 14.5% with Malaise traps only. Among the identified taxa, five are Sardo-Corsican endemics: Bombus terrestris sassaricus, Bombus xanthopus (Kriechbaumer, 1870), Hoplitis legoffi Le Divelec, 2024, Panurgus corsicus Warncke, 1972 and Megachile sicula corsica Benoist, 1935. Six species are reported for the first time from Sardinia: Bombus xanthopus (Fig. 2), Andrena griseobalteata Dours, 1872 (Fig. 3), Colletes brevigena Noskiewicz, 1936 (Fig. 4), Sphecodes crassanus Warncke, 1992 (Fig. 5), Hoplitis praestans (Morawitz, 1893) (Fig. 6) and Hoplitis tridentata (Dufour & Perris, 1840) (Fig. 7). Bee material citations are available in Suppl. material 1, including details such as collection sites, dates, Corine Land Cover classes and associated plant taxa.

Figure 2. 

Bombus xanthopus (Kriechbaumer, 1870) male collected on the Culuccia Peninsula (Sardinia, Italy) A dorsal view B lateral view C head D genital. Scale bars: 2 mm.

Figure 3. 

Andrena griseobalteata Dours, 1872 male collected on the Culuccia Peninsula (Sardinia, Italy) A dorsal view B lateral view C head D genital. Scale bars: 2 mm (A–C); 1 mm (D).

Figure 4. 

Colletes brevigena Noskiewicz, 1936 male collected on the Culuccia Peninsula (Sardinia, Italy) A dorsal view B lateral view C head D genital. Scale bars: 2 mm (A–C); 1 mm (D).

Figure 5. 

Sphecodes crassanus Warncke, 1992 male collected on the Culuccia Peninsula (Sardinia, Italy) A dorsal view B lateral view C head D genital. Scale bars: 2 mm (A, B); 1 mm (C, D).

Figure 6. 

Hoplitis praestans (Morawitz, 1893) male collected on the Culuccia Peninsula (Sardinia, Italy) A dorsal view B lateral view C head D genital. Scale bars: 2 mm (A, B); 1 mm (C, D).

Figure 7. 

Hoplitis tridentata (Dufour & Perris, 1840) female collected on the Culuccia Peninsula (Sardinia, Italy) A dorsal view B lateral view C head. Scale bars: 2 mm.

NJ phylogenetic trees

A total of 212 COI sequences belonging to 61 different species were successfully sequenced (Suppl. material 6: table S4). The NJ phylogenetic tree built with only our new sequences, showed correspondence between monophyletic clades and morphological species, except for Bombus terrestris sassaricus and Bombus xanthopus, which were grouped into a single clade (Suppl. material 2: fig. S1a). In addition, the NJ phylogenetic trees, described below, were built for the following genera and subgenera: Anthophora, Andrena (Chlorandrena, Micrandrena, Taeniandrena), Bombus (Bombus), Colletes, Hoplitis and Lasioglossum (Hemihalictus, Lasioglossum).

Genus Anthophora

Anthophora sp. 1 (MB1_APCUL) clustered with Anthophora atriceps Pérez, 1879 from Spain (but bootstrap < 70; Suppl. material 2: fig. S1b, Fig. 8A), sharing 95.46% similarity (Suppl. material 5: table S3).

Figure 8. 

Neighbor Joining Trees based on the COI of A Anthophora B Colletes C Hoplitis. Taxa marked with a red dot correspond to specimens newly sequenced in this work, while all the others were downloaded from BOLD. Only bootstrap values above 70 are shown at nodes. Collapsed clades correspond to monophyletic taxa (for a non collapsed version of the tree see Suppl. material 2).

Genus Andrena (Chlorandrena, Micrandrena, Taeniandrena)

The two sequences of A. cfr. clypella diverged into two distinct clades without clustering with other species deposited in BOLD (but bootstrap < 70; Suppl. material 2: fig. S1d, Fig. 9A). A. cfr. niveata (GIU9_APCUL) shared 97.81% similarity with its conspecific (Suppl. material 5: table S3), with which it formed a monophyletic clade, but with low support (bootstrap < 70; Suppl. material 2: fig. S1e, Fig. 9B). Samples attributed to Andrena afzeliella were included in the clade formed by conspecifics and specimens of Andrena ovatula (bootstrap > 70; Suppl. material 2: fig. S1f, Fig. 9C), and shared a 99.81–100% similarity with the latter (Suppl. material 5: table S3).

Figure 9. 

Neighbor Joining Trees based on the COI of three subgenera of Andrena A Chlorandrena B Micrandrena C Taeniandrena. Taxa marked with a red dot correspond to specimens newly sequenced in this work, while all the others were downloaded from BOLD. Only bootstrap values above 70 are shown at nodes. Collapsed clades correspond to monophyletic taxa (for a non collapsed version of the tree see Suppl. material 2).

Genus Bombus

Our sequences of Bombus terrestris sassaricus and B. xanthopus formed a monophyletic clade with B. terrestris xanthopus deposited in BOLD (bootstrap > 70; Suppl. material 2: fig. S1c, Fig. 10). However, this resulting group diverged from the only sequence of B. terrestris sassaricus available in the database (but bootstrap < 70; Suppl. material 2: fig. S1c, Fig. 10). Both our samples of B. terrestris sassaricus and B. xanthopus have a high similarity with B. terrestris xanthopus (99.83–100%; Suppl. material 5: table S3).

Figure 10. 

Neighbor Joining Tree based on the COI of Bombus. Taxa marked with a red dot correspond to specimens newly sequenced in this work, while all the others were downloaded from BOLD. Only bootstrap values above 70 are shown at nodes. Collapsed clades correspond to monophyletic taxa (for a non collapsed version of the tree see Suppl. material 2).

Genus Colletes

The sample attributed to Colletes maidli (B38_APCUL) presented 96.95% similarity with specimens of its conspecific from BOLD (Suppl. material 5: table S3). No public sequences are available to confirm such attribution, nonetheless, in the NJ phylogenetic tree our sample was isolated from all the other Colletes species tested (Suppl. material 2: fig. S1g, Fig. 8B).

Genus Hoplitis

In the NJ tree, our sequences of Hoplitis tridentata, H. bihamata and H. praestans diverged from conspecifics (bootstrap > 70; Suppl. material 2: fig. S1h, Fig. 8C), in agreement with the low % of similarity observed in BOLD (96.1%, 94–94.5, 93.65, respectively; Suppl. material 5: table S3). The sequence of Hoplitis legoffi formed a lineage close to Hoplitis adunca (Panzer, 1798), which was its best match on BOLD but with a low % similarity (93.78; Suppl. materials 2, 5: fig. S1h, table S3, Fig. 8C).

Genus Lasioglossum (Hemihalictus, Lasioglossum)

Our sequences attributed to L. leucozonium showed 95.62% of similarity with L. leucozonium specimens available in BOLD (Suppl. material 5: table S3). In the NJ phylogenetic tree, this species was paraphyletic, forming four clades, one of which was occupied by our sequences (Suppl. material 2: fig. S1i, Fig. 11A). Lasioglossum prasinum (GIU11_APCUL) was related to a specimen identified as a conspecific (bootstrap > 70, Suppl. material 2: fig. S1i, Fig. 11A) but their % of similarity was low (93.50, Suppl. material 5: table S3). Our specimens identified as Lasioglossum transitorium planulum clustered with members of this species (bootstrap > 70, Suppl. material 2: fig. S1l, Fig. 11B) with which they shared 99.66% similarity (Suppl. material 5: table S3).

Figure 11. 

Neighbor Joining Trees based on the COI of two subgenera of Lasioglossum A Lasioglossum B Hemihalictus. Taxa marked with a red dot correspond to specimens newly sequenced in this work, while all the others were downloaded from BOLD. Only bootstrap values above 70 are shown at nodes. Collapsed clades correspond to monophyletic taxa (for a non collapsed version of the tree see Suppl. material 2).

Flower visitation network

The interactions between 63 wild bee species and 20 plant families are shown in the flower visitation network (Fig. 12). The plant family visited by the highest number of bee species was Asteraceae (46%), followed by Plumbaginaceae (23.8%), Cistaceae (20.6%), Fabaceae (19%), Asphodelaceae (17.4%), Brassicaceae (15.9%) and Lamiaceae (15.9%) (Fig. 12). In contrast, Apiaceae, Primulaceae and Scrophulariaceae were visited by only one bee species each (Fig. 12). Suppl. material 7: table S5 lists the plant taxa visited by wild bee species.

Figure 12. 

Flower visitation network between wild bee species and plant family on the Culuccia Peninsula (Sardinia).

Faunistic studies are fundamental for understanding species distributions and developing effective conservation strategies. This is particularly important in diversity-hotspot regions with scarce faunal data, such as Sardinia, where information on bee diversity is still limited. Despite the small size of our study area, we recorded six new species for Sardinia, that were previously unreported in the checklist of Italian wild bees (Comba 2019) and in recent faunistic and ecological studies carried out on the island (Catania et al. 2021; Nobile et al. 2021; Lezzeri et al. 2024). Our results emphasize how the knowledge of the occurrence of wild bees is still insufficient and underscore the importance of further research such as in Sardinia. Some of the new records could be explained by the geographical proximity of our study area to Corsica, where the presence of species such as Hoplitis tridentata and H. praestans was already known (Le Divelec et al. 2024) and where Bombus xanthopus is present and considered as endemic to this island, Capraia and Elba (Rasmont et al. 2021). However, further studies may reveal whether these species are restricted to north-eastern Sardinia or are widespread throughout the region.

The bee community sampled in the Culuccia Peninsula included both widely distributed species and taxa with restricted distribution or endemic. Some of these were assessed as being at moderate risk of extinction according to Comba (2019), such as Colletes brevigena, Andrena sardoa, Andrena griseobalteata, Sphecodes crassanus, Panurgus corsicus, Hylaeus deceptorius, Hoplitis bihamata and H. fasciculata. In addition, Lasioglossum prasinum is classified as Near Threatened according to the European and Italian Red List of bees (Nieto et al. 2014; Quaranta et al. 2018).

We recorded several previously unreported bee floral associations, such as Andrena antonellae on Cistaceae, Andrena morio on Plumbaginaceae, Hoplitis praestans on Asteraceae and Andrena afzeliella on Asphodelaceae and Caryophyllaceae (Amiet al. 2004; Cortis et al. 2009; Praz et al. 2022). Moreover, we provide the first observation of floral visits of the Sardo-Corsican endemic Panurgus corsicus, which was found only on Asteraceae species in this study, consistent with the close association with this family already known for other species of this genus (Cross 2020; Wood et al. 2022). The integrative taxonomic approach allowed the species identification of all the samples, except for two Andrena species and one Anthophora species. The latter could not be identified morphologically due to the degraded status of the specimen and its sequence did not show any similarity to any species available on BOLD that would allow a reliable molecular identification. Two Andrena species have not been definitively named due to uncertainties in both their morphological and molecular identification. The genetic divergence between the two sequences of A. cfr. clypella suggests that the samples may represent two different species, despite their strong morphological similarity. Indeed, Chlorandrena species may present aberrant strong divergent sequences within morphologically well-delineated groups (Wood 2023). However, it was not possible to determine the relationship of our specimens with A. clypella, as no public sequences for this species were available in BOLD at the time of the NJ phylogenetic tree analysis (Suppl. material 2: fig. S1d, Fig. 9A). Conversely, the clustering of our A. cfr. niveata sequence with other A. niveata specimens lacks adequate support (Suppl. material 2: fig. S1e, Fig. 9B) and, due to uncertain morphological identification, we have refrained from assigning a definitive name.

In some cases, our samples presented low molecular similarity to their conspecifics (e.g. Lasioglossum leucozonium, L. prasinum and Hoplitis spp.,) but firm identification was provided with the support of specialists after further morphological examination. This low similarity may be attributed to several factors, including insufficient knowledge of genetic variation among populations (i.e., intra-specific diversity), errors in DNA data repositories, and variation in the barcoding gap across all bees (Janko et al. 2024). Additionally, NJ analysis revealed distinct BINs within Lasioglossum leucozonium and L. villosulum, revealing possible cryptic species diversity as previously noted by Villalta et al. (2021) and Janko et al. (2024). A recent integrative taxonomic study employing different molecular analyses identified multiple candidate species within the L. villosulum complex, with morphological analysis resurrecting L. medinai (Vachal, 1895) and L. berberum (Benoist, 1941), while the remaining lineages were identified as L. villosulum (Pauly et al. 2019). Our sequences of L. leucozonium and L. villosulum formed distinct clusters situated between various lineages within these species complexes (Suppl. material 2: fig. S1i, l, Fig. 11A, B), suggesting that the populations in Sardinia may represent a genetically unique entity. The species Hoplitis legoffi was only very recently described by Le Divelec (2024), and its barcode sequence has not yet been deposited in public databases. For this reason, our sample, which was morphologically identified by the taxonomist R. Le Divelec, only matched with H. adunca, though with low similarity, and formed a distinct lineage in the NJ tree (Fig. 8C, Suppl. material 2: S1h).

Another point worth noting is our samples of Bombus terrestris sassaricus and B. xanthopus. These specimens were identified morphologically as two species following Rasmont et al. (2008, 2021) but both showed a high molecular affinity with sequences available in BOLD attributed to B. terrestris xanthopus. Bombus terrestris sassaricus is a subspecies endemic to Sardinia, whereas B. xanthopus was previously considered to be a subspecies of B. terrestris, restricted to Corsica, Capraia and Elba (Estoup 1996; Rasmont et al. 2021). In 2015, the latter subspecies was elevated to species status based on molecular and ecochemical characteristics (Lecocq et al. 2015) and has been considered as such in the recent literature (Rasmont et al. 2021; Ghisbain et al. 2023). However, other studies have provided new evidence that questions the specific status of this taxon. These include molecular analyses (Williams 2021) and the observation of potential hybridization (Boni et al. 2023). In this context, we have relied on morphological classification to identify our samples, pending future studies aimed at clarifying the taxonomy of this diverse and complex genus.

A similar issue was raised from the NJ phylogenetic tree of the subgenus Taeniandrena, in which sequences from BOLD of A. afzeliella and A. ovatula, along with our A. afzeliella specimens, formed a well-supported monophyletic clade (Suppl. material 2: fig. S1f, Fig. 9C). However, sequences labelled as A. ovatula (e.g. Schmidt et al. 2015) were determined prior to the review by Praz et al. (2022), which clarified the distinction between these two taxa based on morphological and molecular differences. This explains the confounding results observed in our study, as these previous misidentifications have not been updated.

This work represents the first study on wild bee species richness on the Culuccia Peninsula and contributes to enrich the currently scarce knowledge on these insects in Sardinia. Furthermore, we added COI sequences to public databases (BOLD and GenBank), contributing to increase the knowledge of the genetic variability of bee species in general and, in particular, for species previously lacking sequences from Sardinia, and often from Italy as well. The integrative taxonomic approach allowed the identification of each specimen more reliably than morphological or molecular identification alone and highlighted potential taxonomic issues. For this reason, further molecular and morphological studies are needed, in particular to achieve accurate species delimitation and contribute to a lower level of understanding of phylogenetic relationships within this diverse group of insects. Finally, this study provides previously undocumented floral associations, highlighting the need for further research into the flower visitation patterns of wild bee species in other areas of the Mediterranean region.

We are sincerely grateful to Achik Dorchin, Thomas James Wood, Michael Kuhlmann, Holger Dathe, Alain Pauly, Romain Le Divelec for their help with the morphological identification of same bee specimens. We also thank Simona Sarmati, Alicia Teresa Rosario Acosta and Alfred Mayer for plant identification. Lucrezia Spagoni is acknowledged for her assistance in the molecular laboratory and Marco Bologna for logistical support in different activities of this work. The field activities were supported by the invaluable help of Paolo Mariottini, Stefano Cantone, Thomas Fusco, Francesca Casale, Lorenzo Fortini and Francesco Forte. Our sincere thanks to Stella Lin Hung and Marco Boglione and all staff of the BIRU S.R.L. Agricola for their excellent support provided in hosting and logistics. This study was supported by the University Roma Tre, Department of Science (grants of Departments of Excellence—L. 232/2016—art.1, commi 314–337 awarded to Dept. of Science—University Roma Tre—Rome—Italy for 2018–2022, and grants 2023–2027). AR is currently supported by the project PON – Ricerca e Innovazione (MUR; Project Code: 999900_PON_RTD_A7-G-15023_SCIENZE). ADG and AR acknowledge the support of NBFC to University of Roma Tre, Department of Science. Funder: Project funded under the National Recovery and Resilience Plan (NRRP), Mission 4 Component 2 Investment 1.4 - Call for tender No. 3138 of 16 December 2021, rectified by Decree n.3175 of 18 December 2021 of Italian Ministry of University and Research funded by the European Union – NextGenerationEU; Award Number: Project code CN_00000033, Concession Decree No. 1034 of 17 June 2022 adopted by the Italian Ministry of University and Research, CUP F83C22000730006, Project title “National Biodiversity Future Center - NBFC”.