Revisions to the faunas of Andrena of the Iberian Peninsula and Morocco with the descriptions of four new species (Hymenoptera: Andrenidae)

Iberia has one of the richest bee faunas in the world, and the genus Andrena is no exception with around 200 species known from the Peninsula. The fauna of Andrena was largely revised in the 1970s, but since then, it has received little attention. Molecular investigation of the taxonomically challenging subgenus Taeniandrena has revealed that the situation is more complicated than previously thought with several cryptic and overlooked species. From the species allied to Andrena (T.) gelriae van der Vecht, 1927, Andrena (T.) gredana Warncke, 1975 stat. nov. from Spain and Portugal is raised to species status, and Andrena (T.) levante Wood & Praz sp. nov. from southeastern Spain is newly described. Furthermore, Andrena (T.) benoisti Wood & Praz sp. nov. is described, having previously been referred to as Andrena (T.) wilkella beaumonti Benoist, 1961. Andrena (T.) beaumonti stat. rev. is itself distinct and restricted to the High Atlas Mountains of Morocco. Outside of the subgenus Taeniandrena, Andrena (Euandrena) fortipunctata Wood sp. nov. and Andrena (Charitandrena) hattorfiana nigricauda Wood subsp. nov. are described from Spain, and Andrena (Notandrena) juliana Wood sp. nov. is described from Spain and Portugal. The male of Andrena (Lepidandrena) baetica Wood, 2020 is also described. Andrena (Euandrena) impressa Warncke, 1967 stat. nov. is raised to species status, displaying a West Mediterranean distribution. Finally, a further two species of Andrena are newly recorded for Spain, Andrena laurivora Warncke, 1974 and Andrena confinis Stoeckhert, 1930. Altogether, these findings reinforce the fact that our understanding of the taxonomy and distribution of Andrena in southern Europe remains incomplete.


Introduction
The Iberian Peninsula is one of the most important places for wild bee diversity globally, with a known fauna of well over 1000 species including many endemics (Ortiz-Sánchez 2011Wood et al. 2020). Its combination of Mediterranean climate, status as a glacial refugium, and wide variety of contrasting habitats (e.g., mountains, scrublands, steppes, and woodlands) supports many different floral communities, thereby offering a large number of distinct foraging niches for wild bee species. Despite a long history of study (e.g., Erichson 1835;Lepeletier 1841;Dours 1873;Pérez 1895;Friese 1897), new bee species continue to be described from Iberia, with the rate of publication actually increasing in recent years (Ortiz-Sánchez et al. 2001;Le Goff 2003;Patiny et al. 2005;Schwarz & Gusenleitner 2010;Müller 2012;Radchenko 2017;Wood & Cross 2017;Kuhlmann & Smit 2018;Le Goff & Gonçalves 2018;Smit 2018;Wood et al. 2020).
The fauna of the Andrena Fabricius, 1775 of Iberia was extensively revised by Klaus Warncke (1967Warncke ( , 1975aWarncke ( , 1976 who described 20 species currently accepted as valid from the Peninsula, including many endemics, as well as a large number of subspecies, some of which may well be valid species in their own right. However, since then, very little taxonomic attention has been paid to Iberian Andrena until recently (Dardón et al. 2014;Wood et al. 2020). Concurrently, molecular genetic techniques have increasingly been applied to identify and resolve outstanding problems in the taxonomy of Andrena (Manderey et al. 2008;Schmidt et al. 2015;Praz et al. 2019). The Iberian fauna of Andrena, with its rich diversity and numerous described subspecies, therefore represents an ideal study system to reveal potential cryptic species with contemporary molecular techniques given the diversity of forms found in this region.
The molecular aspect of this study focused on the monophyletic subgenus A. (Taeniandrena) Hedicke, 1933 , and specifically on taxa allied to Andrena gelriae van der Vecht, 1927. The subgenus itself can easily be recognised in most of Europe because of the broad flattened clypeus that is present in both sexes, and can only nominally be confused with the subgenus Troandrena Warncke, 1975 that is found on the Canary Islands and in the East Mediterranean (Gusenleitner & Schwarz 2002). The two eastern species of Troandrena have a clypeus that is in part clearly shiny (always predominantly dull in Taeniandrena), and the complex of species from the Canary Islands have an integument with clear metallic green tints (always non-metallic in Taeniandrena) and probably belong elsewhere .
Within Andrena (Taeniandrena), A. gelriae was described from the Netherlands. Following the broad taxonomic concept used by Warncke, it has a nominal distribution from Spain, through central and eastern Europe, north to Finland (Gusenleitner & Schwarz 2002). The concept of A. gelriae sensu Warncke included A. gelriae producta Warncke, 1973 which was described from eastern Austria, but this has been treated as a separate species by subsequent authors because of the lack of intermediate forms and distinct male genitalia (Schmid-Egger & Scheuchl 1997;Gusenleitner & Schwarz 2002). Whilst the situation in central Europe is generally clear, things are more challenging in southern Europe as Warncke described the subspecies A. gelriae gredana Warncke, 1975 from Portugal and Spain (Warncke 1975a), and A. gelriae vocifera Warncke, 1975 from southern France (Warncke 1975b). These subspecies have genitalia that markedly differ from the nominate form, but their statuses are unclear and they have not received detailed taxonomic investigation. Because of this lack of investigation, the current distinctions of where to draw the line between the species are somewhat arbitrary, with some publications treating A. gredana, but not A. vocifera, as a valid species (e.g., Nieto et al. 2014).
The present work resolves some of the issues surrounding Iberian Taeniandrena, describing two new species and elevating one species from Iberia and one from Morocco. A further two new Iberian species of Andrena are described, and a West Mediterranean subspecies of Andrena is elevated to species rank. The Spanish fauna of Andrena is also updated to the current level of taxonomic understanding following new discoveries, including a species previously thought to be endemic to Morocco. This work is predominantly based on the examination of previously undetermined Spanish material of Andrena, largely from the Oberösterreichisches Landesmuseum (Austria), and the Naturalis Biodiversity Center, (Netherlands), comprising 2865 female and 1716 male specimens from 159 valid species, as well as investigations into the subgenera Taeniandrena and Euandrena Hedicke, 1933 more broadly at a West Palaearctic scale (Praz et al. 2019).

Material and methods
DNA barcoding was performed on Iberian Andrena (Taeniandrena) to examine the status of the species described here, and on taxa related to Andrena angustior (Kirby, 1802) and A. impressa Warncke, 1967 stat. nov. (subgenus Euandrena). The 658-bp 'barcoding' fragment of the mitochondrial gene cytochrome oxidase I was amplified and sequenced using the primer pair LepF/LepR (Hebert et al. 2004) using the PCR conditions given therein. The primer pair often fails for species of Andrena, possibly due to mismatch in the LepF binding site, resulting in frequent co-amplification and/or low-quality sequences of Wolbachia Hertig, 1936. For this reason, we designed the specific primers COX-Taeniandr-F (TTC TGA CTA YTA CCC CCA TCA A) and COX_Taeniandr_R2 (ATR GGR TCA AAG AAG GAT GA), which together amplify a 365 bp fragment in all species of Taeniandrena examined. Based on the initial comparison of sequences of Taeniandrena, this fragment contains sufficient nucleotide variation for separating all European species of Taeniandrena. For recently collected specimens (< 5 years old), we first tried the combination LepF/LepR, and used the specific primers if the obtained sequences were not clean or if no amplification product was obtained. For older specimens, we used the specific primers mentioned above, and in case these primers did not work, we used the alternate reverse primer COX_Taeniandr_R1 (TGC TCC TAT RAT TGA TGA AAT ACC TG), which amplifies together with COX-Taeniandr-F a 181-bp fragment of CO1. PCR conditions for both specific primer pairs were the following: initial denaturing 94°C for 4 min.; 35-45 cycles of 45 min. at 94°C, 45 min. at 56°C, 45 min. at 72°C; final elongation 72°C for 7 min.
Lab protocols follow Praz et al. (2019); in brief, DNA was extracted from a single leg using DNA extraction kits (Nucleospin, Macherey-Nagel). The PCR products were visualized using agarose gel electrophoresis, purified enzymatically and sequenced with the PCR primers. Chromatograms were trimmed and assembled in Geneious ver. 2020.0.5 and exported consensus were aligned using MAFFT (Katoh & Standley 2013). The resulting matrices were examined and edited in Mesquite (Maddison & Maddison 2018) and converted to amino acid sequences to verify that no stop codon was present. Uncorrected p-distances were computed in a test version of Paup ver. 4.0 kindly provided by D. Swofford (Swofford 2002). Bayesian phylogenetic analyses were performed in BEAST ver. 1.10.4, applying one single partition to which a GTR + G model was applied; the clock model was set to 'uncorrelated relaxed clock' and the tree prior was set to 'speciation: Yule prior'. The analysis was run for 10 million generations, sampling trees and parameters every 1000 generations. The software Tracer ver. 1.7.1 was used to ensure that convergence was reached for every parameter (ESS values > 300). The first 1000 trees were discarded as burn-in and a majority-rule consensus tree was computed with the 9000 remaining trees.
For the analysis of West Mediterranean Taeniandrena, a sequence from Andrena lathyri Alfken, 1899 was used to root the trees, as this species is morphologically divergent from other species of Taeniandrena. In the analysis of Euandrena, a sequence of Andrena fulvago (Christ, 1791) was used to root the tree, following Pisanty et al. (2021). In both cases, a monophyletic group was enforced for all other (ingroup) sequences. Sequences from related taxa were downloaded from BOLD (40 sequences, accession numbers visible in Figs 1-2), maximising the geographic coverage within widely distributed species. For the analysis of Euandrena, several North American members of subgenus Ptilandrena Robertson, 1902 were included, as these are the sister group to the Palaearctic Euandrena . Details on locality information and BOLD accession numbers for all new specimens examined genetically are given in Table 1.
Morphological terminology follows Michener (2007). The abbreviations A, T and S are used for antennal segments, metasomal terga, and metasomal sterna respectively. Specimens were measured from the vertical plane of the front of the head to the tip of the metasoma. Images are presented without scale bars due to their small size and lack of precise submillimetric measuring equipment. Photographs were taken using an Olympus E-M1 Mark II with a 60 mm macro lens and were stacked using Zerene Stacker ver. 1.04 (Zerene Systems, USA) and plates were prepared in GNU Image Manipulation Program (GIMP) ver. 2.10. Post-processing of some images was made in Photoshop Elements (Adobe Systems, USA) in order to improve lighting to highlight specific characters.
Where available, pollen was removed from scopae of female Andrena in order to investigate their dietary preferences following the method of Wood & Roberts (2017). Dietary classification followed Müller & Kuhlmann (2008). Maps were made in R ver. 3.6.0 (R Core Team 2019) using the package ggplot2 (Wickham 2016

Molecular analyses
In the phylogenetic relationships among species of Taeniandrena ( Fig. 1), all species represented by more than one individual formed well-supported monophyletic groups, with the exception of A. wilkella (Kirby, 1802), for which support was low (posterior probability, hereafter PP, of 0.74). These analyses confirm the distinctiveness of A. beaumonti Benoist, 1961 stat. rev. and A. wilkella, which form two separate clades with A. benoisti Wood & Praz sp. nov. sister to A. wilkella and A. beaumonti sister to A. benoisti+A. wilkella. The distinctiveness of A. gelriae s. tr. and A. gredana stat. nov. is also supported; these taxa do not form a monophyletic group. The sampled specimens of A. ovatula (Kirby, 1802) were similar to specimens of A. ovatula s. str. from northern Europe, a taxon that we consider distinct from most populations of 'A. ovatula auct.' from central and southern Europe (Praz & Wood, in prep.). Lastly, A. levante Wood & Praz sp. nov. was the sister species to A. gelriae s. str. (PP less than 0.5).
The analyses of the Ptilandrena-Euandrena (Fig. 2) clade strongly support the recognition of A. impressa stat. nov. as a distinct species and not as a subspecies of A. angustior; these taxa were not closely related in our trees (Fig. 2). The placement of A. impressa (sister to a clade composed of A. fulvata Stoeckhert, 1930, A. angustior, A. allosa Warncke, 1975and A. amieti Praz, Müller & Genoud, 2019 was surprising given that A. impressa shares numerous morphological features with A. fulvata and A. angustior, in particular the broadened gena and long male mandible, this character being absent in A. allosa and A. amieti, displaying the 'typical' unbroadened gena found in Euandrena. Support for this arrangement was however very weak. Average genetic distances within A. angustior and A. fulvata were 0.20% and 0.38%; distances between these two taxa were on average 3.56% (range 3.10-4.15). The distances between the single specimen of A. impressa and these two taxa were considerably higher: on average 8.22% for distances with A. fulvata (range 8.08-8.40) and 8.61% with A. angustior (range 8.39-9.22). With the exception of the difference between A. gelriae and A. levante sp. nov., these between species genetic distances are well above the 2% divergence metric that typically indicates species-level differences (Schmidt et al. 2015).  Warncke, 1975 stat. nov. Figs 3-8 Andrena gelriae gredana Warncke, 1975a: 310 (Spain, Sierra de Guadarrama, ♂).

Remarks
Identification of female material from species close to A. gelriae is highly challenging without access to confidently determined comparative material. In Iberia, female A. gredana can be recognised by the dense punctures of the base of T2 (smooth to weakly and sparsely punctate in A. similis Smith, 1853),  faint punctures on the disc of T1 (strongly punctate in A. wilkella (Kirby, 1802)), fore part of clypeus shiny (Fig. 5), contrasting the central and basal areas which are dull (uniformly dull in other species), the sculpturing of the scutum (Fig. 6), which is shagreened but centrally has a circular area where the shagreenation is comparatively weaker when viewed dorsolaterally (other species uniformly shagreened or with a clearly shiny central circular area), and the comparatively wide hair bands on the terga (Fig. 7) that are complete on T2+3+4 (either much thinner, or not complete on T2 and/or T3), Males are significantly easier to identify, and can initially be recognised by the length ratios of the antennal segments, being part of the group where A3 is equal to A4 in length. In this group, the genitalia of A. gredana are highly distinctive, showing gonocoxa that diverge only slightly at their apexes, and with a penis valve that is clearly narrowed basally and widened to its maximum width centrally before tapering to its apex (Fig. 8). This character allows separation from all other Iberian Taeniandrena. In fact, molecular data shows that this taxon is not closely allied to A. gelriae (Fig. 1), and the divergent structure of the penis valve would support this conclusion when compared to the species allied to A. gelriae (Figs 9-11). Given these differences and its phylogenetic placement, A. gredana is formally raised to species status.

Distribution
Central and northern Spain and Portugal (Warncke 1976 ; Fig. 3). The distribution in Iberia is montane, and it may also occur in the French Pyrenees (see record from Huesca Province; Fig. 3), but no specimens have yet been reported from this region.  (Figs 7, 15). The clearest difference can be seen on the scutum which is strongly and densely shagreened and dull ( Fig. 14), with dense punctures that are almost contiguous (except posteriorly), giving the overall surface a duller appearance than in either A. gelriae or A. gredana ( Fig. 6) that have sparser punctation. The pubescence of the scutum and scutellum is also denser, shorter, and thicker than either comparison species (Figs 4,12).

Andrena (Taeniandrena) levante
As for other similar species of Taeniandrena, male identification is much easier. Males can be recognised as part of the group with A3 equal to or slightly shorter than A4 (A3 1-1.03 times as long as A4). The genitalia are distinctive, with the gonocoxa diverging from close to their base (without their inner margins parallel for at least 50% of their length), the penis valve uniformly wide (not constricted medially), and with the blades of the gonostyli comparatively short, apically as wide as long (Fig. 10). Andrena levante sp. nov. differs from A. gelriae vocifera in numerous morphological features; this taxon will be characterised elsewhere (Praz & Wood, in prep.).

Etymology
The term 'El Levante' is the Spanish name for the eastern part of the Iberian Peninsula that constitutes the majority of the known range of this species (Almería, Granada, Málaga, Murcia, Valencia).
Head. 1.3 times as wide as long (Fig. 13). Clypeus dark, flattened over most of its area, densely and uniformly punctate with exception of raised central impunctate line, punctures separated by < 0.5 puncture diameters, underlying surface shagreened, weakly shining, particularly apically. Face, gena, vertex, and scape with light brownish hair, longest not exceeding half of scape in length. Antennae dark, A4-12 lightened to light brown below. Foveae broad, occupying almost all area between lateral ocellus and top of compound eye, filled with short brown hairs.
MesosoMa. Scutum densely punctate, punctures separated by < 0.5 puncture diameters over majority of surface except becoming slightly sparser centrally and posteriorly, underlying surface shagreened, weakly shining (Fig. 14). Scutellum with sparser punctures separated by 1 puncture diameter, shagreenation weaker, generally shining. Episternum and propodeum with dense raised reticulation, underlying surface dull, propodeal triangle weakly indicated by weak carina, little differentiated from general reticulation. Scutum and scutellum with short, orange-brown, semi-squamiform hairs, episternum with longer light brownish to white hairs, becoming orange-brown on propodeum. Legs dark, hind tibiae and tarsi of mid and hind legs orange, general pubescence light brown basally, becoming orange apically, flocculus, femoral and tibial scopae light brown to golden. Wings hyaline, venation dark orange, stigma orange, nervulus interstitial.
MetasoMa. Terga dark, finely shagreened and weakly shining, apical part of marginal areas lightened semi-translucent brown (Fig. 15). T1 very finely and subtly punctured, punctures on disc scarcely visible against shagreenation, those on margin more visible, separated by 1 puncture diameter. T2-4 more densely and visibly punctate, punctures separated by 0.5 puncture diameters. Terga with whitish hairbands, on T1 represented by two very widely separated spots (separated by almost entire width of tergal margin), T2 widely interrupted, T3+4 complete. Remaining tergal surface covered with short, fine brown to ferruginous hairs visible when viewed obliquely or in profile. Terminal fringe of T5 and hairs flanking pygidial plate golden, pygidial plate rounded, flat, without raised margin.
Head. 1.3 times as wide as long (Fig. 17). Clypeus flattened and densely punctate, punctures separated by < 0.5 puncture diameter, sculpturing as in female. Gena and lower part of face with white hairs, becoming light brown on scape and vertex, longest equalling length of scape. Antennae dark, A4-13 extensively lightened to dark brown below. A3 as long as A4.
MesosoMa. Scutum, scutellum, episternum, and propodeum structurally as in female (Fig. 18). Scutum and scutellum with fine light brown to golden hairs that equal length of scape, becoming light brown to whitish on propodeum and episternum. Legs dark, apical tarsal segments lightened dark red, pubescence whitish to light brownish. Wings hyaline, venation dark orange, nervulus slightly postfurcal.

Remarks
In his original description, Benoist (1961) drew a comparison between A. beaumonti stat. rev. and Andrena flavipes Panzer, 1799, presumably because of the dense hair bands that help characterise this species. Warncke (1967) noted that this association was clearly incorrect at a subgeneric level, associating the bee with members of the subgenus Taeniandrena instead. On the basis of its large and dense puncturing, its golden hairs flanking the pygidial plate, and the orange metatarsi of the second pair of legs, Warncke (1967) associated this taxon with Andrena wilkella, but as a subspecies because of the unbroken hair bands on T3+4. He then went on to report A. wilkella beaumonti from Iberia (Warncke 1976; see also Ortiz-Sánchez 2011, giving it a distribution of Morocco, Spain, and Portugal (see also Lhomme et al. 2020). However, genetic barcoding shows that A. beaumonti is distinct, and Moroccan material is also morphologically different from Warncke's concept of A. wilkella beaumonti in Iberia, most obviously by lacking a shiny spot on the scutum that is well differentiated from the surrounding shagreenation. It can also be separated by the finer hairs on the scutum, the generally paler pubescence, and by the thicker and more complete hair bands on terga 2-4 (Figs 20-27), meaning that A. wilkella beaumonti sensu Warncke in Iberia is actually undescribed (see below). The previously undescribed male of A. beaumonti can also be distinguished using the same characters (colour of pubescence, nature of tergal hair bands) and also by the length of A4 which is 1.15 times as long as A3, whereas in A. benoisti Wood & Praz sp. nov. (see below) it is 1.3 times longer (Figs 30-31). There are also slight differences in the genitalia, with a comparatively narrower penis valve and more clearly separated gonocoxa (Figs 34-35) in A. beaumonti.

Distribution
Morocco, from the High Atlas Mountains in the area south of Marrakesh. The village of Asni is found at a moderate elevation of 1200 m. It is not clear exactly where Arroud is because of changes in spelling practices, but it probably refers to the village Aroumd some 25 km SSE of Asni. Contemporary sampling locations of A. beaumonti stat. rev. are relatively close to Asni itself. Tizi-n-Test is some 60 km SW of Asni, and Oukaimeden is even closer, just 10 km to the SE. The site of Oukaimeden is at a much higher elevation of 3200 m, and the cluster of records in this region suggests that A. beaumonti is restricted to high-elevation sites flying in May and June.  (Fig. 26, thin and widely interrupted on T2+3 in A. wilkella), and the sculpturing of the scutum, which is shagreened except for the centre where it has a circular shining area that strongly contrasts with the remaining surface which is dull (Fig. 24).
The males can be recognised by the ratio of the antennal segments where A3 is clearly shorter than A4 in length (Fig. 30), the terga are strongly and densely punctate, the tergal margins are depressed with a thin, shiny, puncture-free apical zone, and T2-4 have thick hairbands, medially interrupted on T2, complete on T3+4 (Fig. 32). They are therefore closest to A. wilkella, but the thicker and more complete hair bands in combination with its larger body size (10-11 mm against 8-9 mm) allows for differentiation. Genetically, A. benoisti sp. nov. is placed as the sister of A. wilkella, a position that is corroborated by their morphological similarity.

Etymology
Named after Raymond Benoist, the French botanist and entomologist who described Andrena beaumonti stat. rev. from Morocco, this name being incorrectly applied at the subspecific level to the taxon we describe here.
Head. 1.3 times as wide as long (Fig. 22). Clypeus flattened, densely but shallowly and weakly punctured, punctures separated by 0.5 puncture diameters, faint impunctate line visible centrally, underlying surface shagreened, dull, slightly shining apically. Lower face and gena with whitish hairs, becoming light brown to golden on scape, upper face, and vertex, hairs not exceeding length of scape. Antennae dark, A5-12 slightly lightened grey below. Foveae broad, occupying almost all area between lateral ocellus and top of compound eye, filled with short brown hairs.
MesosoMa. Scutum densely punctured over most of surface, underlying surface dull, punctures separated by 1 puncture diameter except becoming sparser centrally and posteriorly forming circular shape, here punctures separated by 2 puncture diameters, underling surface weakly shagreened, shining, contrasting remaining surface (Fig. 24). Scutellum shiny, densely punctate centrally, separated by 0.5 puncture diameters, punctures sparser laterally. Episternum with slightly raised reticulation, underlying surface dull, reticulation reaches lateral faces of propodeum but weakly, disappearing before propodeal triangle, this marked by change in surface sculpturing, internal surface with fine, sparse, and slightly raised reticulation. Scutum and scutellum with short, orange-brown, semi-squamiform hairs, episternum with longer light brownish to white hairs, becoming orange-brown on propodeum. Legs dark, hind basitarsi fully and hind tibiae partially orange, general pubescence light brown. Flocculus and femoral scopa whitish, tibial scopa golden. Wings hyaline, venation and stigma golden, nervulus postfurcal.
Head. 1.3 times as wide as long. Clypeus flattened, densely punctate, punctures separated by 0.5 puncture diameters with exception of clear longitudinal impunctate line, underlying surface shiny. Gena, face, vertex, and scape with long whitish to light brownish hairs, equalling or exceeding scape in length. Antennae dark, A3 1.3 times as long as A4 (Fig. 30).
MesosoMa. Scutum, scutellum, episternum, and propodeum structurally as in female, though propodeal reticulation a little more extensive. All parts of mesosoma with long light brown to golden hairs, equalling or exceeding scape in length. Legs dark, Hind tibiae entirely and hind tarsi apically coloured dark orange, pubescence whitish to light brown. Wings hyaline, venation dark orange, nervulus postfurcal.

Distribution
From the Sistema Central around Madrid westwards through the Sierra de Gredos into central Portugal, and north into northern Portugal, Galicia, and Zamora (Fig. 3). Both A. benoisti sp. nov. and A. gredana stat. nov. show a remarkably similar distribution and occurring at many of the same localities (see also Warncke 1976), particularly in the Sistema Central from which true A. wilkella seems to be absent.  Warncke, 1975).

Etymology
The name 'forti' ('strong') + 'punctata' ('punctured') was chosen because of the pronounced punctures visible in both sexes that help separate this taxon from other Iberian Euandrena.
Head. 1.2 times as wide as long (Fig. 38). Clypeus weakly arched, very densely punctate, punctures separated by < 0.5 puncture diameters to almost contiguous, underlying surface shiny. Gena, face, and scape with black hairs, face between antennal insertions with few grey hairs, vertex with intermixed black and golden hairs, hairs at most equalling length of scape. Antennae dark, A4-12 lightened grey below. Fovea narrow, occupying ⅓ of area between lateral ocellus and compound eye, comma shaped, filled with black hairs.
MetasoMa. Terga dark, margins strongly depressed and lightened semi-translucent brown (Fig. 39). Terga densely and uniformly punctate, punctures separated by 0.5-1 puncture diameters, punctures not extending onto depressed tergal margins, excluding margins underlying tergal surface shagreened, weakly shining. T2-4 with hair fringes arising from junction between disc and margin, fringes of sparse white hairs overlying marginal areas, not obscuring underlying surface. Terminal fringe of T5 and hairs flanking pygidial plate dark brown, pygidial plate with centrally raised longitudinal area and raised apical rim, weakly punctate, dull.
MesosoMa. Structurally as in female, entire surface with long light brown to golden hairs, exceeding scape in length. Legs dark, tarsi lightened, dark red apically, pubescence golden. Wings hyaline, venation and stigma orange, nervulus prefurcal.

Remarks
The subgenus Euandrena is taxonomically challenging, and it is likely that we are just scratching the surface of hidden species diversity in southern European mountain chains (Praz et al. 2019). The Andrena of the Iberian alpine areas have received little recent attention, and within this context, the presence of an undetected alpine Euandrena in the high mountains of Spain is less surprising, but emphasises how much more there is to learn about European diversity of Andrena even in a comparatively well-studied region.

Distribution
The Sierra de Gredos in the Sistema Central to the Cantabrian mountains in northwestern Spain in the province of Asturias. 52,54,56,58,60,62,64 Diagnosis Andrena hattorfiana (Fabricius, 1775) can be very easily recognised as it is the only known member of the Charitandrena Hedicke, 1933 in the West Palaearctic. Females have a unique combination of a conspicuously carinate hind femur, a hind tibial spur that is convexly broadened towards the middle, and an elongate glossa. Males are harder to recognise through a combination of unique characters, but their large body size, yellow clypeus, lack of a pronotal carina, and long glossa in combination with specific characters such as genital structure are sufficient to facilitate identification.

Andrena (Charitandrena) hattorfiana nigricauda
Andrena hattorfiana nigricauda Wood subsp. nov. differs from the nominate form (Figs 51,53,55) because it has an almost completely darkened integument and pilosity. True Andrena hattorfiana normally shows extensive variation in the colour of its abdominal integument, with individuals from the same location varying from extensively red marked to completely black. Its variability has resulted in a relative large number of synonyms for an Andrena (Gusenleitner & Schwarz 2002). However, what is consistent is that the hairs of T5 and those flanking the pygidial plate are uniformly golden-orange (Fig. 55). In the Spanish material presented here, these hairs are completely black (Fig. 54). In the male, there are no clear structural differences, but as in the female the pubescence is much darker across the whole body (Figs 58-63).

Etymology
The name 'nigri' ('black') + 'cauda' ('tail') was chosen to reflect the black hairs at the apex of the metasoma in the female sex, a clear point of difference from the nominate form.
Head. Black, 1.2 times as wide as long (Fig. 52). Clypeus domed, slightly flattened centrally and apically, underlying surface shagreened and dull basally and laterally, becoming weaker apically, weakly shining. Entire surface densely punctured with exception of faint impunctate central line, punctures separated by 0.5-1 puncture diameters. Process of labrum triangular, twice as broad as long, deeply emarginate apically. Gena, vertex, face, and scape with moderately long dark brown to black hairs, longest equalling length of scape. Antennae dark, A3 long, exceeding length of A4+5+6, A5-12 lightened orange below. Fovea of a uniform width, occupying half of area between lateral ocelli and top of compound eye. Vertex wide, as wide as three ocellar diameters.
MesosoMa. Scutum and scutellum finely shagreened, weakly shining, densely and evenly punctured, punctures separated by 1 puncture diameter. Episternum and propodeum evenly and finely rugose, propodeal triangle clearly marked by a lateral carina, internal surface with more strongly produced rugosity, broadly similar. Scutum and scutellum with short, episternum and propodeum with longer dark brown hair, at its longest not exceeding length of scape. Legs dark, apical tarsal segments lightened red-brown, pubescence dark brown to dark red. All scopal hairs dark brown, those of femoral and tibial scopa ventrally finely plumose, otherwise simple. Wings weakly infuscate, venation dark brown, nervulus interstitial.
Head. Black, 1.2 times as wide as long (Fig. 60). Characters as in female, but clypeus yellow with exception of two dark spots laterally, A3 slightly subequal to A4+5+6, and A4-13 lightened orange below.

Remarks
The presence of A. hattorfiana nigricauda Wood subsp. nov. in the mountains of Alicante represents, at the very least, an isolated population showing a unique colour pattern. Other records of A. hattorfiana in Spain come from the Sistema Central around and to the west of Madrid, the Picos de Europa, and the Pyrenees (Gusenleitner & Schwarz 2002). The location in Alicante is therefore around 400 km from the nearest known populations of A. hattorfiana that conform to the typical hair colour pattern. However, separating this population as specifically distinct is premature without molecular investigation, and overall differences are slight outside of colouration; the male genitalia of the two colour forms are essentially indistinguishable. Andrena hattorfiana nigricauda subsp. nov. could simply be a melanic form, and molecular analysis could show whether it is nested within A. hattorfiana when considered across its whole range. Indeed, material of A. hattorfiana from the most southerly parts of the Balkan Peninsula has hairs flanking the pygidial plate that are dark brown, but otherwise this material shows typical red terga (treated as A. h. dimidiata Brullé, 1832 by Warncke, see Gusenleitner & Schwarz 2002).

Distribution
Andrena hattorfiana is distributed from central and northern Spain across Europe to Greece, Turkey, and the Caucasus (Gusenleitner & Schwarz 2002).
Head. Black, 1.1 times as wide as long (Fig. 67). Clypeus black, slightly domed and slightly flattened centrally, densely punctured laterally with punctures separated by 0.5 puncture diameters, becoming sparser centrally, separated by 1 puncture diameter, narrow central impunctate line present. Gena, clypeus, and inter-antennal area with white hairs, vertex with a mixture of white, brown, and black hairs, and frons and inner margin with black hairs, all hairs variable in length, longest exceeding length of scape. Antennae uniformly dark, A3 slightly shorter than A4+5.

Remarks
Andrena baetica is morphologically close to A. mocsaryi Schmiedeknecht, 1883, but differs in the shorter, squamous hairs on the scutum (Wood et al. 2020). The discovery of this male material further confirms the proximity of A. baetica to A. mocsaryi through genital structure (Fig. 69, see illustrations in Schmid-Egger & Scheuchl 1997), but also confirms its specific status as it possesses an entirely black clypeus, whereas A. mocsaryi has a yellow clypeus, and the apex of S8 is truncate to slightly emarginate, whereas in A. mocsaryi it is slightly pointed (Schmid-Egger & Scheuchl 1997).

Diagnosis
This taxon can be placed in the subgenus Notandrena in the female sex because of its broad head (clearly broader than wide), the weakly rugose (not shagreened) propodeal triangle, the clearly punctured metasoma, and by the dorsolateral angle of the pronotum with a weak transverse ridge. In the male sex, recognition is easy because of the greatly enlarged and carinate gena in combination with the broadened apex of the gonostyli.
Eight species of the subgenus Notandrena are known from Iberia (Wood et al. 2020). Female Andrena juliana Wood sp. nov. can be recognised in the group of A. nitidiuscula Schenck, 1853 because of its small size (7-8 mm, excluding A. erythrocnemis Morawitz, 1870, A. langadensis Warncke, 1965, andA. urdula Warncke, 1965 which average 11-12 mm in length) and dark hind tibiae (orange in A. chrysosceles (Kirby, 1802)). Within this group, it lacks the distinctive shortened dorsal scopal hairs of A. pallitarsis Pérez, 1903, the gena and vertex are normal (vertex clearly less than the diameter of a lateral ocelli, clearly greater than this distance in A. foeniculae Wood, 2020), and there is no deeply impressed line on the front half of the scutum (clearly impressed in A. nitidiuscula). In an Iberian context, it is therefore closest to A. fulvicornis Schenck, 1853 (alternative character state in parentheses), but differs in the sparser punctures on T1, separated by 2-3 puncture diameters (separated by one puncture diameter, Fig. 77), by the shinier scutellum (scutellum shagreened and clearly dull, Fig. 73 Greece, and the European part of Turkey (Gusenleitner & Schwarz 2002;Schwenninger 2013). It can be separated by the sculpturing of the clypeus which is shiny centrally (centrally shagreened) and by the sculpturing of the scutellum which is shiny (strongly shagreened and dull).
Males can be quickly recognised by their yellow clypeus in combination with their small size (7 mm), a character that is also found in A. chrysosceles and A. pallitarsis (A. erythrocnemis, A. langadensis, and A. urdula with the clypeus yellow marked, but larger, averaging 10 mm). In both these species, the yellow markings cover the entire clypeus and extend onto the lower paraocular areas, whereas in A. juliana sp. nov. this marking is restricted to the centre of the clypeus. The gena is noticeably more carinate, and as in the female the scutum is also noticeably shinier. The genitalia conform to the typical shape found in members of the nitidiuscula group ( Fig. 83; see illustrations in Schmid-Egger & Scheuchl 1997;Schwenninger 2013). As for other members of this group, the gonocoxa are weakly shagreened in a similar fashion to that found in the Zonandrena Hedicke, 1933. However, the outer margin of the gonostyli is almost straight, whereas there is a clearer obtuse angle here in A. chrysosceles, A. pallitarsis, and A. fulvicornis (see illustrations in Schmid-Egger & Scheuchl 1997;Schwenninger 2013). As in the female sex, the male is also similar to the eastern A. curvana. However, apart from the yellow clypeus (black in A. curvana), the genitalia are different, with less pronounced gonocoxal teeth, and lacking the pronounced impressed channel in the basal section of the gonostyli (see illustration of this character in Schwenninger 2013).

Etymology
The name is taken from the locus typicus, San Julián.
Head. Dark, broad, 1.4 times as wide as long (Fig. 71). Clypeus domed, laterally and dorsally shagreened and densely punctate, punctures separated by 0.5 puncture diameters, centrally shiny and less densely punctate, punctures separated by 0.5-1.5 puncture diameters, weakly-marked impunctate line present. Process of labrum small, rectangular, front margin very weakly emarginate. Gena as wide as compound eye, gena, vertex, and face with whitish hairs, these not exceeding length of scape. Antennae dark, A5-12 lightened orange below, A3 equalling A4+5, both A4 and A5 clearly broader than long. Facial fovea neither wide nor narrow, occupying half space between lateral ocellus and compound eye. Ocelloccipital distance short, less than half diameter of lateral ocellus.
MesosoMa. Scutum weakly shagreened, gently shining, densely but shallowly punctured, punctures separated by 0.5-1 puncture diameter centrally (Fig. 72). Scutum in fore half with barely impressed mid line. Scutellum more clearly shiny, punctures dense only at margin and forming a central longitudinal line, otherwise sparse and separated by 3-4 puncture diameters. Episternum and propodeum finely reticulate, dull, propodeal triangle barely indicated by very fine carina, internal structure essentially unchanged. Scutum and scutellum with short whitish pubescence through which longer hairs protrude, these approaching length of scape. Episternum and propodeum with longer whitish hairs, longest equalling length of scape. Legs dark, apical tarsal segments and apex of basitarsi coloured orange, this extending slightly onto apex of hind tibiae. Wings hyaline, venation orange, nervulus interstitial.

Remarks
Using the key to Iberian Notandrena of Wood et al. (2020), females of A. juliana Wood sp. nov. key to couplet seven which separates A. fulvicornis and A. nitidiuscula. It can be separated from A. nitidiuscula by the absence of an impressed longitudinal line on the front half of the scutum, and from A. fulvicornis by the less dense punctures of T1 and the shinier scutellum. No male key was produced because the male of A. foeniculae is unknown, but males of A. juliana sp. nov. should be identifiable by the combination of their genital structure and yellow clypeus. Additionally, A. foeniculae flies in August and September, and so based on the limited number of specimens collected to date, both taxa are unlikely to fly together at the same time. The similar species A. curvana and A. fulvicornis are bivoltine and fly in both the spring and the summer. Since A. juliana sp. nov. is known only from spring material, further study is needed to clarify its full period of activity.

Andrena (Euandrena) impressa
sequence closely matched sequences from northern European populations, and the phylogeny suggests that A. impressa is more strongly differentiated than A. angustior and A. fulvata are from each other (Fig. 2). Though the molecular phylogeny suggests that A. impressa is more distantly related to A. angustior+A. fulvata than A. allosa+A. amieti, support for this placement is weak, and based on the morphology of the male head we consider A. angustior+A. fulvata+A. impressa to form a trio of related but distinct species.
The diet of A. angustior in Britain was documented by Wood & Roberts (2017 A. angustior is a species of Atlantic woodland edges, utilising spring flowering herbs and shrubs but interestingly not the trees themselves. Fewer data are available for Andrena impressa; the two analysed pollen loads contained Cistaceae (50.3%, Cistus), Scrophulariaceae Juss. (42.9%, Scrophularia L.), and Asteraceae Bercht. & J.Presl (6.9%, Anthemis-type). Given its distribution, a diet more focused on Mediterranean herbs and shrubs is expected, but more study is required.

Distribution
The distribution map of Warncke (see Gusenleitner & Schwarz 2002) is broadly correct in the northern distribution of A. angustior, the continental distribution of A. fulvata, and the West Mediterranean distribution of A. impressa (France, Spain, Portugal, Algeria, Morocco), but incorrect in the omission of true A. angustior from cooler parts of Iberia (Fig. 84). This distribution indicates that A. angustior is not simply a northern species but rather one with an Atlantic affinity. Though we did not examine material from this region as part of this study, A. angustior is likely to be present in cooler parts of western

Remarks
Material from southwestern Spain closely resembles Moroccan specimens, but are generally less hirsute, particularly in the single male Spanish specimen, although this may be simply due to the condition of the individual specimen (Figs 101-108). Structurally, there are no clear differences, and so they are considered to be conspecific. The pronouncedly Atlantic distribution of Morocco and southwestern Iberia is not unprecedented in bees, as the extremely rare Lasioglossum musculoides Ebmer, 1974 has a global distribution of the Souss Valley in southwestern Morocco and southwestern Portugal (Pauly 2015). Coastal areas of Huelva and Sevilla continue to produce records of range-restricted species Figs 101-108. Andrena laurivora Warncke, 1974 (TJWC). 101, 103, 105, 107 taxa are present in mountainous parts of northern Spain, though given the limited material available it is difficult to come to firm conclusions concerning their true distributions.

Distribution
The distribution of A. confinis is unclear because of continued confusion and synonymy with Andrena congruens Schmiedeknecht, 1884. At this moment, it is not possible to give an accurate picture of its true distribution, but both species are present in central Europe, with A. congruens having more southern and A. confinis more northern tendencies (Schmid-Egger & Scheuchl 1997).

Discussion
In total, this investigation into the fauna of Andrena of Iberia has identified three good taxa deserving a taxonomic elevation to species status, and four previously overlooked or cryptic species. Combined with the discovery of A. laurivora, previously thought to be endemic to Morocco, these findings comprise a total of seven species newly recorded for both the Spanish and broader European fauna, three for the Portuguese fauna, and the replacement of A. wilkella with the endemic A. beaumonti stat. rev. in Morocco.
In common with other recent investigations into the Andrena of central and southern Europe (Praz et al. 2019;Wood et al. 2020), targeted taxonomic attention has revealed a considerable complexity that was not sufficiently resolved by previous authors. This can partly be explained by philosophical differences. The leading West Palaearctic taxonomist of Andrena Klaus Warncke described much of the diversity he encountered at the subspecific level, in the case of this present study the taxa A. gredana stat. nov. and A. impressa stat. nov., or nested other good taxa within a broad species concept in the case of A. beaumonti stat. rev. As our power to distinguish between distinct lineages with molecular genetics has increased, our understanding of how Andrena should be classified at the species level has also changed, as what could be considered to simply be 'variable' lineages can be identified as distinct. As a result, the subspecific classifications of Warncke that are so prevalent in his work on the Andrena of the Mediterranean basin will continue to be revised (Praz et al. 2019), likely resulting in a substantial increase in species richness for the genus.
In addition to these philosophical changes, the Iberian Peninsula itself genuinely harbours an overlooked diversity. The major Iberian revision of Andrena by Warncke (1975a;1976) was naturally limited by the material available for study, a large part of which came from the collection of Dusmet y Alonso who collected predominantly around Madrid. As such, much of the diversity of Andrena present in the eastern part of the Sistema Central and the region of La Mancha south of Madrid was described in those publications, but there was relatively little material available to study from mountains in Andalucía or in the northwestern part of the Peninsula. The absence of material from northwestern Iberia can be illustrated in the panurgine genus Flavipanurgus Warncke, 1972. The southern Iberian species F. ibericus (Warncke, 1972) was described well before its northwestern Iberian sister species F. kastiliensis (Warncke, 1987) after more material for study became available (Warncke 1972(Warncke , 1987; elevated to species status by Cross & Wood 2018). A more recent study has revealed yet more hidden diversity of Andrena in this region (Wood et al. 2020), and these northwestern mountains, from the western part of the Sistema Central to northern Portugal and the Cantabrian, Léon, and Galician mountains host an endemic andrenid fauna of A. benoisti Wood & Praz sp. nov., A. fortipunctata Wood sp. nov., A. omnilaevis Wood, 2020, F. kastiliensis, and the majority of the population of A. gredana stat. nov., as well as serving as a transition zone where typically Atlantic or central European species such as A. angustior and A. wilkella can be found close to their Iberian or Mediterranean counterparts.
The second major area not well covered by Warncke's revisions were mountains in southern Iberia, and more broadly Andalucía itself as Dusmet collected relatively little from this region and Warncke did not personally visit these areas either, and many of the endemic species they support have only been recently described (e.g., Müller 2012). As such, it should be less of a surprise that they host an endemic diversity of Andrena such as A. levante Wood & Praz sp. nov. What is more surprising is the presence of newly described taxa in southern Iberia that are close to taxa more typically found in central and eastern Europe, but which are separated by substantial distributional gaps. Wood et al. (2020) described A. baetica which is very close to the eastern A. mocsaryi Schmiedeknecht, 1883, and Kuhlmann & Smit (2018) described Colletes jansmiti Kuhlmann, 2018 which is very close to Colletes punctatus Mocsáry, 1877, the nearest records of both comparison species being found in eastern Austria. Andrena juliana sp. nov. can broadly be added to this collection because of its similarity to A. curvana, another typically eastern species. With limited material available for morphological or molecular study, all these taxa require further investigation to better understand their specific biogeographical situations and whether these are driven by consistent underlying trends.
Finally, studies of the subgenus A. (Taeniandrena) are far from complete, even in Iberia. We did not investigate the problems surrounding the A. ovatula (Kirby, 1802) complex that potentially contains up to three species in Iberia, and an unclear number across the entire Mediterranean basin. What is clear is, that there is a great deal of cryptic diversity present in southern European Andrena which is remarkable given that the European bee fauna is the best studied bee fauna globally. Given the discoveries made here, it illustrates the quantity of work needed to identify and delineate bee species in less well-studied parts of the world.