The 5,908 wild bee specimens registered in the MNHNL collection were distributed amongst 33 genera from six families (Andrenidae, Apidae, Colletidae, Halictidae, Megachilidae and Melittidae). Exact sampling sites and detailed information for each museum specimen are available at the Species Observation Database Service of the museum. From the 218 wild bee species listed in the database, 194 were present in previous species checklists, confirming that at least some individuals have been found in Luxembourgish territory. The remaining 24 species did not figure as present in the country in the literature sources consulted. These species were represented in the collection exclusively by individuals collected in other countries, except for five species, whose annotations linked them to at least one voucher specimen from a gathering site placed in modern Luxembourgish territory: Colletes fodiens (Fourcroy, 1785), Hylaeus styriacus (Förster, 1871), Nomada furva (Panzer, 1798), Nomada sexfasciata (Panzer, 1799) and Hoplitis papaveris (Latreille, 1799). In three of these cases, previous taxonomic annotations of the analysed voucher specimens and our morphological re-examinations were in agreement (for N. furva, N. sexfasciata and H. papaveris).

However, in the remaining two cases, the taxonomic annotations showed conflicting results. In the case of C. fodiens, there were two registered specimens (one male and one female) in the collection. The results of the morphological re-evaluation of both specimens were inconclusive. The female specimen (MNHNL41840), which was collected in 1999, lacked the conspicuous white-haired face and those hairs across the dorsal surface of the first tergite that characterise C. fodiens (sensu Amiet et al. 1999, Falk and Lewington 2015 and Falk and Lewington 2015). Some authors mention that C. fodiens females have a tendency to lose hair as they age, which might make them resemble C. similis females (Falk and Lewington 2015). However, without these traits, C. similis, as a potential annotation, cannot be excluded. The male specimen (MNHNL41841), which was collected in 2013, had the hairs of the sternites glued together and partially covered by sand grains. The genitalia were contracted inside the abdomen, obscuring visualisation of the crucial characters described in the taxonomic keys. As a more conservative approach, we thus do not recognise C. fodiens as a potential new finding for Luxembourg.

In the case of H. styriacus, there were eleven female specimens in the collection. In all cases, the shape of the facial fovea matched the one of Hylaeus from the subgenus Paraprosopis (sensu Dathe et al. 2016) to which H. styriacus belongs. However, the morphological re-evaluation of the voucher specimens suggested different annotations. One of the specimens (MNHNL41982) presented the tridentate mandible and pronotum with pointed edges that characterise Hylaeus clypearis (Schenck, 1853). In the rest of the cases evaluated, the characteristics of the specimens matched the description of Hylaeus pictipes (Nylander, 1852): facial lateral yellow spots, punctures on tergites/mesopleura and the overall shape of the head. For the aforementioned reasons, the original annotation and the presence of H. styriacus in Luxembourg remains to be confirmed.

Finally, the DNA barcodes of two of the specimens from the pilot phase of the wild bee atlas project had a 100% genetic similarity in BOLD with sequences annotated as Andrena lagopus (Latreille, 1809) and Sphecodes majalis (Pérez, 1903), respectively. Both species have no previous records in the country, rendering those two entries the first documented findings.

Therefore, the morphological and molecular evidence indicates the presence of five wild bee species that are missing in the most current checklists (Suppl. material 1)

Although, in most cases, the visual inspection of the collection specimens confirmed the results of previous taxonomic assignments, there were two instances of disagreement in which the original taxonomic annotation could not be confirmed with our morphological re-evaluation.

In the case of the potential C. fodiens registered in the collection, traditional taxonomic techniques did not allow us to confirm the suggested annotation due to unfavourable specimen conditions. Even though the morphological assessment provided inconclusive results, molecular taxonomic tools might be able to provide further information regarding the identity of similar Colletes specimens. Given that the two closely-related species C. fodiens and C. similis have an estimated genetic distance of 2.17% (Schmidt et al. 2015), DNA barcoding may be effective to separate the two taxa in fresh specimens. Regarding the Hylaeus individuals which were originally described as H. styriacus, the morphological keys, used in the re-evaluation, suggested two different annotations for the bees stored in the collection (H. clypearis and H. pictipes). It is worth mentioning that the most important morphological traits used to separate species within the subgenus Paraprosopis, according to the key of Dathe et al. (2016), could only be observed by using a large-scale free-angle observation system, which outperforms the capacity of a common light microscope. Due to the size of the Hylaeus bees (~ 4 mm), distinguishing similar species without such a level of magnification and optical performance might not be straightforward, as it may be hard to observe the most subtle details in the head and face. The importance of this factor as a potential error source in the morphological identification of Hylaeus bees and other wild bees of similar size remains to be determined.

The two new findings of Andrena lagopus and Sphecodes majalis require a special ecological discussion. Both were collected in 2019 during the pilot phase of a wild bee atlas project. In particular, the discovery of the ground nesting bee, A. lagopus, is locally relevant. This species is described as an oligolectic, warm-habitat-loving bee with an Atlanto-Mediterranean distribution (Westrich and Schwenninger 1997, Zettel et al. 2019). Given its specialised preference for pollen sources from the family Brassicaceae, the dispersal and distribution of this species is strongly dependent on the presence of its plant hosts, such as Brassica napus (most common pollen source), Sinapsis arvensis, Barbarea vulgaris and Cardamine pratensis (Westrich and Schwenninger 1997). Although the foraging habitats of this bee most commonly include rapeseed fields, set-aside arable fields, ruderal sites, orchards, meadows and gardens, A. lagopus has been described as a pioneer species with a high capacity for colonising new foraging habitats (Westrich and Schwenninger 1997). Once considered an endangered species in Germany, it is hypothesised that the large-scale cultivation of rapeseed might have contributed to its recovery and spread (Westrich and Schwenninger 1997, Zettel et al. 2019). The Luxembourgish specimen was found in a meadow in Kehlen, where plants from the Brassicaceae family might be available nearby. Moreover, the specimen presented fully loaded scopae when it was collected, indicating that foraging activity took place shortly before sampling, probably in the surrounding area. Further research would be necessary to determine whether the newly-recorded presence of A. lagopus in Luxembourg is related to a recent expansion of its plant hosts. As the pollen loads were collected and preserved in glycerol, future pollen analysis might provide further information on this matter.

Another interesting finding comprises the discovery of S. majalis in an orchard in Manternach. This cuckoo bee has a Western Palaearctic distribution that extends across Spain, Algeria, Belgium, southern France, Italy, Hungary, Romania, Switzerland, southern Ukraine, Croatia and The Netherlands, with a potential subspecies (Sphecodes majalis barbatus) in Turkey (Warncke 1992, Raemakers 2004). It is considered one of the rarest blood bees in Central Europe and very little is known about its biological cycle (Herrmann et al. 2003, Raemakers 2004, Streese 2020). Available information suggests that its life cycle and parasite-host relationship might be quite unique: it is a species-specific brood parasite of Lasioglossum pallens, a warm-climate-loving sweat bee with a preference for trees as pollen sources (Herrmann et al. 2003). It has been speculated that the narrow flight season of S. majalis (males typically fly from mid-April to early May and females from mid-April to late May) might be a consequence of the particular relationship this cuckoo bee has with its host. Unlike most other blood bees, which visit different nests and kill or expel their occupants, it has been observed that S. majalis females do not drive away its host, showing no aggressive or defensive behaviour towards L. pallens females (Herrmann et al. 2003). As the cuckoo bees seem to be tolerated by its hosts, it is possible that the female S. majalis remains in the nest for much longer periods than other blood bees, reducing the chances of spotting them in the field (Herrmann et al. 2003). Most likely, this high level of specialisation constrains the distribution of S. majalis to the occurrence of its host bee, as well as the types of habitat in which it can be found. So far, L. pallens has not been reported in Luxembourg, but it has been described as present in nearby countries, such as France and Belgium (Rasmont et al. 2017). It is expected that further sampling campaigns will reveal its presence in the country as well.

Since S. majalis and L. pallens demonstrate particular nesting and foraging behaviour, their apparent rarity might be an artefact of inadequate sampling techniques (Herrmann et al. 2003). However, it cannot be ruled out that local changes in land and biotope management might actually have led to an increase in the number of suitable habitats for these two wild bee species, contributing to their expansion into climatically-favourable areas (Kitt and Reder 2014). For example, it has been hypothesised that the finding of both species in Belgium and the Netherlands (seemingly outside their known geographical range) might correspond to a recent expansion event (Raemakers 2004). The species may have entered the region from the south-west (through France), in a dispersal pattern that excludes the Rhine Valley (Raemakers 2004). Nevertheless, this hypothesis is challenged by findings of both species in the German Federal State of Rhineland-Palatinate (Kitt and Reder 2014) and even at locations as far north as the Botanical Garden of the University of Potsdam (Streese 2020). Future investigations could determine whether the finding of S. majalis in Luxembourg supports any of these hypotheses regarding potential dispersal routes and mechanisms, but more specimens would be needed to provide proper arguments for this debate.

Our results highlight the importance of NHCs as sources for discoveries and critical re-interpretations of scientific knowledge. In combination with recent fieldwork material, four new wild bee species (A. lagopus, H. papaveris, N. furva and S. majalis) were added to the national checklist of Luxembourg. Additionally, we found evidence that supports the current presence in the country of a fifth species (N. sexfasciata), which was omitted during past inventories. With the addition of these findings, the number of wild bee species registered in Luxembourg has increased to 346. The wild bee collection curated at the National Museum of Natural History Luxembourg is preserved under suitable conditions, which will allow future generations of researchers to use, re-examine and debate it in order to answer scientific questions pertinent to their own historical time. As such, our results represent just a small fraction of the exceptional value that NHCs have as repositories for the documentation of the bio- and geodiversity of the world. Despite this great potential, the current decrease in the available funds to keep and curate NCHs threatens the future of several collections, especially the smaller ones, affecting our possibilities to continue profiting from them in the years to come. Therefore, support from the scientific community and funding bodies is imperative, so biological collections can keep growing, documenting and ultimately, fulfilling their role in society.