EUROPEAN JOURNAL OF ENTOMOLOGY EUROPEAN JOURNAL OF ENTOMOLOGY The bee fauna (Hymenoptera: Apoidea: Anthophila) of allotments in downtown Lisbon

. In the last ten years, a growing number of studies have focused on urban green areas as potential refuges for biodiversity, where private gardens, urban parks and green roofs have relatively high diversities of wild bees. However, the western Mediterranean is still poorly studied and is a biodiversity hotspot that is already suffering the consequences of climate change. It is essential to rectify this and understand how urban settings can support biodiversity. In this context, this study provides an assessment of the taxonomic and functional composition of bee assemblages in three allotments in downtown Lisbon, Portugal. Using only an entomological net, we collected 202 specimens from April to July 2018, belonging to ﬁ ve families, 20 genera and 58 species, of which six are rare species in Portugal and nine ﬁ rst records for the Lisbon district. Megachilidae was the most diverse family, comprising 15 species, while Apidae was the most abundant family. Most of the species identi ﬁ ed were solitary and had a generalist pollen diet, with a low incidence of social and parasitic species.


INTRODUCTION
It is widely recognized that pollinators and the ecosystem services they provide are under increasing pressure, with both climate change and land-use conversion considered to be the drivers globally (Goulson et al., 2015;Mazor et al., 2018). In Europe, livestock breeding, pesticides and habitat loss and fragmentation are the primary reasons for the decline in wild bees (Rundlöf et al., 2015;Bongaarts, 2019;Sirami et al., 2019;Senapathi et al., 2021).
Pollination is an essential ecosystem service, with bees considered one of the most important pollinators (Bates et al., 2011;Winfree et al., 2011). They are responsible for supporting the production of a wide variety of foods, mainly vegetable and fruit crops, and keystone species in both human managed and natural ecosystems (Kleijn et al., 2015). It is known that pollinators have a signifi cant economic contribution to crop production (Gallai et al., 2009;Kleijn et al., 2015). Unfortunately, of the 2,000 species in Europe, 9% are threatened with extinction and for 57% there is not enough data to assess their risk of extinction (Nieto et al., 2014).
Bees are functionally described based on their life-history traits (food, sociality and nesting requirements). Some taxa display fl oral specifi city (oligolectic), while others are pollen generalists (polylectic) (Dötterl & Vereecken, 2010). The 20,000 species described globally are either so-recorded for each transect, focusing only on species richness. After collection, bees were individually stored in tubes fi lled with ethanol 70% and kept in a freezer (-20°C) before being processed, mounted and identifi ed to species level using several identifi cation keys (Amiet et al., 2001(Amiet et al., , 2002(Amiet et al., , 2007(Amiet et al., , 2010(Amiet et al., , 2014Michener, 2007;BWARS, 2016;Michez et al., 2019;Asher & Pickering, 2020) and the reference collection of ISA (Instituto Superior de Agronomia, Universidade de Lisboa). The identifi cation of specimens was carried out with the aid of a binocular magnifying glass (NexiusZoom) in the laboratory. Of the 215 specimens collected, it was not possible to identify 13 individuals to species level due to damage. When it was impossible to identify to species level, we sent the specimens to taxonomists specializing in European bees. The identifi cation of the fl owering plants was done using the Flora-On platform (Portuguese Botanic Society, 2014). At all three allotments there were no honeybee hives or commercial bumblebee colonies.

Compilation of the ecological traits of bees
Information in the literature on their ecological traits was compiled (Michener, 2007;Matteson et al., 2008;Baldock et al., 2015Baldock et al., , 2018Kratschmer et al., 2019;Michez et al., 2019;Lanner et al., 2020) and when no information was available we contacted the bee specialists. Species were described based on their life-history traits (LHTs), dividing each trait into two categories: nesting (below-ground or above ground), sociality (solitary or eusocial) and pollen preference (polylectic or oligolectic) (Westrich, 2019). Parasitic species were placed in a third category as they do not build nests or collect pollen for their brood and lay their eggs in nests of their hosts (Bogusch et al., 2006;Bogusch & Straka, 2012).

Statistical analysis
The number of genera, families and relative frequency of the species captured were determined. The relative frequency of LHTs was also measured. The same analysis was performed for the fl owering plants. To address diversity, Hill numbers with three different exponents were calculated: (q) of diversity with a q value of 0 for species richness with all species having the same weight; q = 1 for the exponential of Shannon's index with species being weighted according to their abundance in the community and q = 2 for the inverse of the Simpson's index, which favours abundant species (Jost, 2006;Tuomisto, 2010). The differences between the values of the Hill numbers gives the degree of unevenness or dominance in each community (Jost, 2006). The higher the dominance in a community, the bigger the differences between these three values.
As this study included many species for which only one or two individuals (singletons and doubletons) were recorded, the sampling effort was determined using a species accumulation curve of the number of species by randomly selecting the order of each transect. To determine total species richness (sampling on all dates pooled) in the allotments sampled, non-parametric estimators Chao1 and Jackknife1 (both abundance based) were used (Picanço et al., 2017). These reduce the under-sampling effect that infl uences the number of species recorded (Sobs) (Walther & Moore, 2005;Fetridge et al., 2008;Russo et al., 2015).

Bee fauna
In total, 202 specimens were included in the analysis, made up of fi ve families, 20 genera and 58 species (Table  1). The species Xylocopa violacea (Linnaeus, 1758) and Bombus terrestris (Linnaeus, 1758) were recorded in the al., 2015;Hofmann & Renner, 2020;Lanner et al., 2020), whereas for the Mediterranean region there is a clear lack of such studies. It is crucial to address this as the same conservation measures cannot be applied in every region due to differences in their environments, land and geographical history (Aguirre-Gutiérrez et al., 2015Connelly et al., 2015;Mallinger et al., 2016;Durant & Otto, 2019;Grab et al., 2019;Bogusch et al., 2020).
The Mediterranean area provided perfect conditions for bee speciation, which resulted in a high diversity of bees (Michener, 1979;Nielsen et al., 2011;Tscheulin et al., 2011;Baños-Picón et al., 2013;Pisanty & Mandelik, 2015;Sanchez et al., 2020). In Europe, bee species richness increases from north to south with the highest species richness in the Mediterranean region. Besides having a higher species richness, the Iberian, Italian and Balkan peninsulas also have the highest concentrations of endemism (Nieto et al., 2014), which is similarly associated with a general north-south positive gradient. Two main factors can explain these patterns: (i) a more favourable energy and water balance in the Mediterranean areas, which has resulted in an extremely high fl oral diversity (Petanidou & Ellis, 1997;Potts et al., 2003;Sébastien et al., 2009); (ii) the likely role these areas had as refuges during the Quaternary glaciations (Feliner, 2011).
This study aims to improve our understanding of the diversity of wild bees in urban areas in a species-rich, but poorly studied, Mediterranean biodiversity hotspot. It is crucial to understand bee communities and how green areas in cities can promote wild bee conservation. The possibility of declines in bees due to changes in fl oral availability associated with climate change makes it especially important to determine how cities can help in conserving bees. Therefore, this study aimed to answer the following questions: (1) Which wild bee communities are attracted to allotments in downtown Lisbon, a highly urbanized Mediterranean city? and (2) What are the ecological requirements of these species?

Study location
In recent years, Lisbon has increased the number of managed green areas, such as allotments, mainly associated with urban parks. This study was conducted at three different allotments in downtown Lisbon (38°44´N, 9°8´W) ( Fig. 1): Quinta da Granja (8400 m 2 ), Quinta das Flores (5200 m 2 ) and Quinta Conde D'Arcos (4500 m 2 ), from April to July 2018. These sites are in similarly highly urbanized environments and the surrounding fl ora is managed by Lisbon city hall. These allotments are mainly used for urban agriculture and organically farmed for growing similar species of fl owering plants. The three gardens are at least 2 km from each other (beyond the foraging distances of bees).

Species sampled
Wild bees were collected using an entomological net. Sampling took place between 10 am and 4 pm to maximize captures, on warm and sunny days (15-30°C), with low wind speed and no rain. In each allotment, 15 transects (20 m × 3 m) were each scanned for 20 min, capturing any bees on vegetation, ground, structures such as fences or in fl ight. Plant species in fl ower were three allotments, identifi ed in the fi eld and then released. Therefore, they were excluded from the abundance analysis. The European honeybee, Apis mellifera Linnaeus, 1758, was also recorded at all sites but not captured.
The family Megachilidae was the most diverse family, with a total of 15 species, followed by Apidae, Andrenidae, Halictidae and Collectidae. Regarding the number of species in each genus, Andrena had the highest species richness, comprising 11 species, followed by Hylaeus with eight species. For the genera Amegilla, Ammobates, Bombus, Colletes, Xylocopa and Sphecodes only one species was recorded, whereas the highest number of specimens (35 specimens) was recorded for Eucera. The data collected included some rare species (one or two records). In total, of the 58 species captured, 17 (29%) were based on only one (singletons) and 14 (24%) on two individuals (doubletons). The species, Lasioglossum malachurum (Kirby, 1802) and Eucera elongatula Vachal, 1907, were the most abundant, with 17 and 14 individuals respectively, making up around 15% of the total assemblage.
The species accumulation curve based on sampling effort indicates a non-stabilization of bee species richness across sites (Fig. 2). Based on Hill numbers signifi cantly fewer species (degree of unevenness or dominance in the community) are associated with the last than the fi rst number [q = 0 (58.0); q = 1 (38.9); q = 2 (28.8)], indicating around 50% fewer species of bees. Depending on the estimators used (Chao 1 and Jackknife 1) and for all three locations combined the potential specifi c richness is between 66 and 77 species, which is greater than that recorded in the present study (58) (Fig. 2, Table S1).

DISCUSSION
The Mediterranean area lacks studies on wild bees in urban settings, although this group has attracted attention in recent years. For Portugal, a study focused on the diversity of pollinating insects in small-scale farming landscapes in the Oeste region (2012-2013) reported about 70 species of bees belonging to 16 genera (Catarina Reis, pers. commun.). A two-year study (2014)(2015) in the north of the country in which the abundance and diversity of bees at sites at different stages of land abandonment were compared, reports 157 species of bees, with several the fi rst confi rmed records for Portugal, based on capturing 2721 specimens (Penad o et al., 2022). In Lisbon, a study carried out at Tapada da Ajuda in Lisbon's biggest green area, recorded 32 species in allotments in 2016 (Carvajal et al., 2017).
As referred to above, there are studies on small-scale farming landscapes and semi-natural/natural landscapes, which are two of the most extensive land-use classes in Portugal. To our knowledge, this is the fi rst study to focus on the diversity of wild bees in the Mediterranean area in urban areas, the third main land-use class in this country (CLMS, 2 018).
In the present study, 58 species were recorded in three allotments in Lisbon city centre, which is approximately 8% of the 712 species listed as occurring on the Portuguese mainland ( (Table  1). It is worth stressing that 37 species recorded in this study are not reported by Carvajal et al. (2017). Hereafter, a total of 69 species are now reported as occurring in allotments in downtown Lisbon.
The highest species richness was recorded for the Megachilidae, which may be linked to their fl exibility in nest selection as they are known to use several types of aboveground cavities in human-made structures (Matteson et al., 2008;MacIvor & Packer, 2015;Egerer et al., 2020). On the other hand, although Andrena was the most diverse genus (11 species), the Andrenidae were the least abundant (10% of total records). The abundance of these bees differs from that reported in other studies in central Europe (Banasz ak-Cibicka & Żmihorski, 2012;Geslin et al., 2015;Sirohi et al., 2015), where the Andrenidae are the most abundant. The low incidence of these bees might be due to the fact that this family primarily consists of ground-nesting species that forage in spring, which in 2018 was late and it rained in the early months (IPMA, 2018). In urban areas, in the Mediterranean region, they are usually associated with grassland and ruderal habitats, where the soil is rarely cultivated and there are more native species of plants (Banaszak-Cibicka & Żmihorski, 2012). The Apidae were the most abundant (31% of total records), probably because it is the most diverse family in terms of morphology and behaviour (Michener, 2007) and as a consequence resilient and adaptable in their response to changes in their habitat. The most dominant species, Lasioglossum malachurum (8% of total records), is a eusocial species present in spring  and summer, which possibly accounts for the high number of specimens collected. The genus Lasioglossum is also referred to as one of the most resilient to land-use change mostly due to its life histories traits (Grab e t al., 2019).
There are several European based studies showing the potential of urban areas for supporting a high species richness of wild bees (Banaszak-Cibicka & Żmihorski, 2012;Geslin et al., 2015;Sirohi et al., 2015;Langellotto, 2017;Banaszak-Cibicka et al., 2018;Buchholz et al., 2020;Lanner et al., 2020;Theodorou et al., 2020). In the present study, a high species richness was recorded for a very small assemblage, indicating that the number of species should be higher than that recorded. The accumulation curve also showed that the number of species did not stabilize throughout each season. It is expected that the species richness would increase between seasons once there is a natural succession of spring and summer species (Michez et al., 2019). However, in cities like Lisbon, with hot summers and green areas subject to high maintenance (removal of wildfl owers), the succession could be negatively affected as fewer fl owers are available (Hamblin et al., 2018). Regarding the diversity indexes (q = 0, 1, 2), the community has a high degree of unevenness (two dominant species, Lasioglossum malachurum and Eucera elongatula), which is explained by the loss of almost 50% of species richness from the fi rst to the last Hill numbers. This loss is probably due to a high number of singletons and doubletons, indicating an under-sampled assemblage. Our estimators (Chao 1 & Jackknife 1) corroborate this as they predict between 66 and 77 species at the sites sampled. Although there are several studies on wild bee richness in urban areas (Geslin et al., 2015;Sirohi et al., 2015;Buchholz et al., 2020;Lanner et al., 2020), they differ in the percentage of species reported in terms of each country's total bee fauna. Comparing our results with those of the above studies pose problems due to different sampling methods and their primary focus. In addition, each city has a different spatial planning and geographic setting leading to different bee assemblages (Sirohi et al., 2015). Similarly, studies focusing on bee LHTs instead of taxonomic diversity have a better potential for making comparisons and drawing better hypotheses/conclusions (Sheffi eld et al., 2013). On the other hand, compared with studies in the Mediterranean countryside (Pisanty et al., 2015;Hevia et al., 2016;Rodrigo Gómez et al., 2021;Penado et al., 2022), the number of species in urban settings should be lower, as was recorded here.
In this study most of the wild bees were solitary (85%), pollen generalists (81%) and had similar percentages in both nesting requirements (48% below-ground or 44% above-ground). These fi ndings are comparable to those of studies in Poland and Austria, the fi rst in urban public gardens (Banaszak-Cibicka & Żmihorski, 2012) and second in communal gardens (Lanner et al., 2020), although a lower number of social species were recorded in this study. Mediterranean regions generally have a low percentage of social species in their faunas because of radically different hot summers (Hamblin et al., 2018). Also, in allotments there are few areas of bare soil in such small patches used for the intensive production of crops, making it almost impossible for the right conditions to exist for large colonies of eusocial species to persist. Most of the bees collected were polylectic. As in previous studies, urban areas tend to have a low incidence of oligolectic species (Matteson et al., 2008;Banaszak-Cibicka & Żmihorski, 2012;Geslin et al., 2015). Although there were few oligolectic species they do occur in allotments. Of the 11 oligolectic species recorded, their preferences ranged from Asteraceae, Brassicaceae, Fabaceae, Boraginaceae to Ericaceae (Table S2), which matched the incidence of these fl ower taxa in the transects. It is common to have similar plants in gardens as many exotic fl owers are planted by local gardeners. This could potentially mean that by planting specifi c fl owers from different taxa, other specialist bees could thrive in allotments (Langellotto, 2017;Egerer et al., 2020). Almost half of the species captured nest in cavities above-ground. Cavity nesters do not rely on patches of sparsely vegetated ground. They use many different cavities from cracks in walls, cavities dug by other insects or themselves, hollow stems or tree trunks (Matteson et al., 2008;Fortel et al., 2014). It is also widespread to use reeds or bamboo for garden fences and to grow plants, which bees use for nesting and feeding (MacIvor & Packer, 2015). Therefore, the sites sampled hosted a high richness of megachilids (15 species). Parasitic wild bees made up 6.9% (four species) of the species of bees collected, but only 3.5% of the specimens. These bees are indicators of the stability of wild bee communities (Sheffi eld et al., 2013), as they tend to respond very quickly to habitat disturbances (Fortel et al., 2014). In the present study, the low number of parasitic species could be due to the absence of specifi c bee hosts and/or low number of samples.
In conclusion, this study showed that allotments can potentially host a diverse wild bee fauna in a highly urbanized Mediterranean city. These green islands provide food and nesting places for bee communities, favouring solitary species and pollen generalists. On the other hand, a few eusocial, parasitic and pollen specialist species were also recorded. The actual richness is likely to be higher than that recorded, due to the very low number of samples. Nevertheless, it is worthy of mention that in the small assemblage (202 specimens) recorded there were 58 species of wild bees. In addition, some bees like the megachilids occur more frequently than other bees in allotments. So, specifi c conservation measures like increasing the availability of nests (bee hotels) and planting a variety of endemic plants could enhance the "refuge potential" of these areas since they are watered all year and as a consequence remain favourable habitats during hot and dry summers. It is also essential to study non-urbanized areas in order to better understand the importance of these habitats as a baseline for direct comparison with urban settings.
In the future the use of different sampling methods should increase the number of species and individuals captured and enable more different types of locations to be sampled so that an accurate plan of action can be devel-oped for Lisbon allotments. Although this study included only three allotments, nevertheless, it shows they are important in urban areas in the Mediterranean area. In addition, the effect of different types of urban planning should be investigated, since the level of land heterogeneity and geographical history infl uence the wild bee fauna. Given the lack of studies on wild bee diversity in the Western area of the Mediterranean, coupled with the growing threats to these insects, the results presented increase our are level of understanding, but a great deal more research is needed.

AUTHORS' CONTRIBUTIONS:
MA did the experiments, analysed the data, prepared fi gures and tables; EF and MTR conceived and designed the experiments; All authors were involved in the writing, reviewed drafts and approved the fi nal draft of this paper.

CONFLICTS OF INTEREST:
The authors declare that they have no confl ict of interest.