Euandrena specimens with black and orange pubescence can be found infrequently across north-western Africa. They have typically been referred to as A. bicolor in the literature, but two distinct taxa are present. One corresponds to A. bicolor s.l., but the other is clearly distinct genetically; morphologically it can be distinguished by the structure of the clypeus, with A. bicolor s.l. with the clypeus shiny between the punctures, whereas in the second taxon the puncture interspaces are shagreened and dull, and there is a weak longitudinal furrow, similar to what can be seen in species like A. (Euandrena) angustior, though more apically situated and occupying a shorter distance.

Warncke (1974) used subspecific concepts for dealing with A. bicolor, with two subspecies used for North African material – A. b. oraniensis Lepeletier, 1841 and A. b. agraria Warncke, 1974 (see also distribution maps in Gusenleitner and Schwarz 2002). However, the use of these names is not immediately straightforward. Andrena oraniensis Lepeletier, 1841 was described from Oran in northern Algeria (Lepeletier 1841) and is a confused and unclear taxon. In the MNHN collection, the type series cannot be located, and it may have been lost as many specimens have been moved between different boxes by past workers and curators. In the Pérez collection, there are long series of Andrena (Melandrena) florentina Magretti, 1883 that were identified by Pérez as A. oraniensis. Warncke (1967) noted this, but argued that this must be incorrect due to size, and instead placed the name in combination with A. bicolor as a subspecies found in North Africa. Warncke (1974) later authored the replacement name A. bicolor agraria Warncke, 1974 for Andrena (Euandrena) nigriventris Pérez, 1902 which was described from northern Algeria and Morocco (locus typicus Tangier, following lectotype designation) but which is a preoccupied name, nec. Apis nigriventris Gmelin, 1790 which is a synonym of A. (Melandrena) nitida (Müller, 1776). Warncke then used both oraniensis and agraria as subspecies, giving an overlapping distribution in Warncke (1974, e.g. both occurring at Oukaimeden in the High Atlas), but in his distribution maps (see Gusenleitner and Schwarz 2002) giving a non-overlapping distribution, with oraniensis more or less occurring only in Algeria and Tunisia and agraria occurring only in Morocco.

In the original description of A. oraniensis, Lepeletier (1841: 245) draws attention to the colour of the hairs on the hind legs. Specifically, he states that: “cuisses des deux postérieures garnies des poils ferrugineux pales ; leurs jambes et leurs tarses à poils noirs en dessus, ferrugineux en dessous”. This bicoloured tibial scopa (dark dorsally, ferruginous ventrally) does not correspond at all to members of the A. bicolor s.l. group which have uniformly orange tibial scopae. It does however correspond very well to the concept of A. florentina which has a distinctive bicoloured scopa, one of the characters that allows it separation from its sister taxon A. bicolorata (Rossi, 1790). Based on the identifications made by Pérez, who probably saw the original specimen in Lepeletier’s collection and Lepeletier’s original description, the position is taken that A. oraniensis cannot be a Euandrena taxon displaying the colour pattern of A. bicolor s.l. However, without a type, it is undesirable to make A. oraniensis the senior synonym for A. florentina. Andrena oraniensis is therefore declared a nomen dubium until such time as the original syntypic series can be located.

The next oldest available name from North Africa is A. bicolor agraria. Pérez (1902, as A. nigriventris) writes: “Abdomen très luisant, particulièrement les depressions, que précèdent les bourrelets très prononcés”. This is suggestive, as the distinct North African taxon has strongly depressed tergal margins. However, barcoded A. bicolor s.l. from Morocco also show this character, so it is not diagnostic in and of itself. Examination of the lectotype specimen of A. nigriventris (designated by Warncke 1967; Fig. 6) shows that the clypeus is apically smooth and shing between the punctures (Fig. 6C), meaning that it cannot belong to the distinct taxon, and it remains a synonym of A. bicolor s.l. One additional name is available, that of A. (Euandrena) fervida Pérez, 1902, a taxon described from Algeria. The lectotype is in poor condition, lacking its metasoma (Fig. 7). Diagnosis is therefore challenging, but the clypeus is shiny between the punctures and it cannot be conspecific with the distinct North African taxon. The position of Warncke (1967) that this name is synonymous with A. bicolor s.l. is maintained. The distinct North African taxon would therefore appear to be undescribed. No taxonomic action is taken in the current work – further genetic data will either demonstrate a closer affinity with the specimen from the Sierra Nevada, in which case this name will be available to apply to North African populations, or these sequences will confirm its distinct nature, and it can be described. In any case, a name is required for the Iberian lineage which is unambiguously undescribed.

Material examined. Andrena bicolor: Algeria: Theniet El Had [35.8727°N, 2.0007°E] (1♀, MNHN (lectotype of A. fervida); Morocco: Tanger [35.7537°N, -5.7906°W], 1♀, MNHN (lectotype of A. nigriventris Pérez).

Andrena (Euandrena) lavandulae Pérez, 1902: 156 ♀♂ [France, lectotype by present designation: MNHN].

Andrena (Euandrena) angustior impressa Warncke, 1967: 234, ♀♂ [Morocco: OÖLM, examined] syn. nov.

Remarks. Warncke (1967) synonymised A. lavandulae with A. bicolor without apparently inspecting the type. This is not explicitly clear in the text of Warncke (1967), but in this publication he typically reported lectotypes in the Paris collection when he designated them, and usually labelled paratypes as well, often taking a selection for his personal collection. There are no specimens of A. lavandulae in the Warncke collection, which combined with the lack of mention of a lectotype in Warncke (1967) leads me to believe that he did not inspect the type series. He then described Andrena angustior impressa Warncke, 1967 in the same publication, a taxon that was subsequently elevated to species status by Wood et al. (2021) based on genetic and morphological evidence.

In the MNHN collection, specimens of A. lavandulae are labelled as lectotype and paralectotype by Teunissen (August 1986; Fig. 8), but these designations were never published. The female specimen is therefore recognised here as a lectotype, by present designation. Examination of these specimens shows that they are clearly the same taxon as A. impressa and distinct from A. bicolor, with the weakly depressed, shagreened, and weakly shing tergal margins, the black terminal fringe and hairs flanking the pygidial plate, the intermixed light and dark hairs on the face, and in the male sex by the broadened gena (broader than the width of the compound eye). The lectotype comes from Banyuls-sur-Mer in the south-west of France, which falls within the distribution of the taxon as presented by Wood et al. (2021). Additional genetic sequences from Spain and Morocco closely match the Portuguese sequence of A. impressa presented by Wood et al. (2021), forming a well-supported clade (Fig. 5; bootstrap support of 93) that is distinct from both A. angustior and A. bicolor. Andrena impressa syn. nov. is therefore synonymised with A. lavandulae sp. resurr. The distribution is Morocco, Algeria, Portugal, Spain, and France (Wood et al. 2021).

Material examined. France: Banyuls [Banyuls-sur-Mer, 42.5658°N, 2.8658°E], 1♂, 1♀, MNHN (female lectotype, by present designation).

Andrena fuscosa was described from southern Spain by Erichson (1835). Andrena fuscosa has had a complicated taxonomic history, and has been described many times due to its variation in colouration and size (Gusenleitner and Schwarz 2002). Iberian genetic data are therefore of interest since the peninsula represents the locus typicus. Iberian sequences were identical with each other, and closely matched sequences from Israel and India (the latter identified as A. (Plastandrena) agilissima Scopoli, 1770 but clearly misidentified; Melanapis was actually described as a genus based on material from India), being separated by an average of 1.53% (range 0.88–1.82%; Fig. 9). They were more clearly separated from Moroccan sequences, by an average of 5.43% (range 3.94–5.93%). Given the lack of morphological differences between North African and Iberian specimens, a broad interpretation of this species is taken here. Including all sequences together, average intraspecific distance is 3.14% (range 0.00–6.23%) which is considered to be acceptably small. Additional study is necessary to establish whether North African material is consistently distinct; the first nominally available name would be A. (Melanapis) rutila Spinola, 1838 which was described from Egypt.

Andrena (Melandrena) morio Brullé, 1832: 353, ♀♂ [Greece: MNHN, not examined].

Andrena (Melandrena) hispania Warncke, 1967: 212, ♀♂ [Spain: OÖLM, examined] syn. nov.

Remarks. Andrena hispania Warncke, 1967 was described from Algeciras in southern Spain. The identification characters given by Warncke are comparatively weak and rely on hair colour and the degree of infuscation of the wings, without mentioning definitive structural characters. Genetically, the three A. hispania sequences from Spain and Portugal mixed with A. morio sequences from Israel, Morocco, Portugal, Spain, Tunisia, and Turkey without forming a cluster, this group having bootstrap support of 100 (Fig. 10). Confusingly, additional sequences from Greece, Morocco, and Tunisia formed two sister groups to this clade. There is no clear geographic pattern, and so a broad A. morio concept is adopted, including A. hispania syn. nov. as a direct synonym of A. morio. Additional genetic data using more powerful techniques are required before an alternative taxonomic conclusion can be drawn. Andrena morio is known to be highly variable in its colouration, leading to a higher than average number of synonyms for an Andrena species (Gusenleitner and Schwarz 2002). Andrena hispania was classically thought of to have only a single generation in the spring, which may explain why its wings are less infuscate than A. morio, particularly in the summer generation of this species that can be noticeably darker than spring flying individuals.

Remarks. The specific differences between these taxa is unclear across southern Europe. In some northern countries, only two taxa are present (A. nitida and A. thoracica, e.g. the United Kingdom), with no introgression observed. Andrena nitida flies only in the spring, whereas A. thoracica is bivoltine, flying in both the spring and the summer. In Central Europe, A. limata can be found, this taxon also being bivoltine. Differentiation between the three taxa in Central Europe has often utilised hair colouration characters, as in this region the three taxa are separable with reference to the hairs on the sides of the mesosoma (light in A. nitida and A. limata, dark in A. thoracica) and the hairs of the face and the tibial scopal (light and dark in A. nitida, uniformly dark in A. limata and A. thoracica). There are additional characters such as the colour of the hind tibial spur and the density of the punctures T1, but these are less commonly referred to; both A. nitida and A. limata have dense punctures on T1 (separated by up to 1 puncture diameter), whereas they are more clearly spaced in A. thoracica (punctures separated by 1–2 puncture diameters). It is important to note that A. limata is a replacement name for A. lucida Lepeletier, 1841 nec. A. lucida Panzer, 1798 which is nominally a synonym of A. bicolor Fabricius, 1775 but this must be established (see section on subgenus Euandrena). Andrena lucida Lepeletier was described from ‘France’, without further details. I have not been able to inspect the type which should be in the MNHN collection, but which I could not locate.

The situation in southern Europe is much more challenging. In south-western Europe, A. limata becomes much darker, and therefore closely resembles the colour form of A. thoracica, with extensive black pubescence on the mesosoma laterally. This colour form was described as A. limata mixtura Warncke, 1967 (illustrated by Wood et al. 2020a). This form can be recognised by the dense punctures of T1, separating it from Iberia A. thoracica; the colour of the hind tibial spur is variable and of little consistent utility. Because of this variation, Warncke later adopted a much broader concept of A. nitida, with A. nitida s. str., A. nitida limata, and A. nitida mixtura (e.g. Warncke 1974). This position was not followed by Gusenleitner and Schwarz (2002) who followed a three distinct species interpretation (nitida, limata including mixtura, thoracica).

Analysis of barcodes does not provide clarity (Fig. 10). Whilst A. thoracica and A. nitida form reciprocally monophyletic clades, three different clades are formed for A. limata. There is no geographic pattern to these clades; clade #1 contains individuals from Austria, Poland (KJ837115; specimen identified as A. nitida but associated photograph on BOLD shows A. limata colour pattern), Spain, and Portugal; clade #2 contains individuals from Morocco, Spain, and Portugal; and clade #3 contains individuals from Finland (MZ625969; specimen identified as A. nitida but associated photograph on BOLD shows A. limata colour pattern), France, and Spain. Confusingly, sequences IBIHM501-21 and WPATW086-21 come from two different female specimens collected from the same field near Pedret i Marzà in north-eastern Spain. As Iberian individuals fall into three of these clades and the power of COI analysis is clearly insufficient to deal with potential hybridisation and introgression, no further action can be taken on the basis of these results. A dedicated study using more powerful genetic markers is necessary to resolve this issue. For now, I maintain the position of Gusenleitner and Schwarz (2002) in that there are three species in this group (nitida, limata including mixtura, and thoracica). It will probably be necessary for a future study to designate a neotype for A. limata if Lepeletier’s original specimens cannot be definitively located.

Andrena (Melandrena) discors Erichson, 1841: 192, ♀ [Algeria: ZMHB, examined].

Andrena (Melandrena) creberrima Pérez, 1895: 46, ♀♂ [Algeria: MNHN, examined] syn. nov.

Andrena (Melandrena) vachali Pérez, 1895: ♀ [Tunisia, lectotype by present designation: MNHN] syn. nov.

Andrena (Melandrena) bodemeyeri Benoist, 1969: 246, ♀ [Tunisia: MNHN, examined].

Remarks. There has been extensive confusion between A. discors and A. creberrima. Erichson (1841) described A. discors from Algeria; a specimen is preserved in the ZMHB collection which may be automatically the holotype, but this is ambiguous as it is unclear if Erichson described the species from multiple specimens (Fig. 11). In any case, the specimen is a syntype, and Erichson described the species only in the female sex. The specimen conforms to the classical concept of the species, having the typical features of the former members of the subgenus Zonandrena, i.e. a clypeus with punctures forming a pattern of weak longitudinal channels. The pubescence of the face is black (Fig. 11C), and there are numerous intermixed dark hairs on the scutum (Fig. 11D). The tibial scopae are orange (Fig. 11E), and there are weak indications of light tergal hair bands but these are degraded and scarcely visible (Fig. 11F). Warncke (1974) gave numerous localities for A. discors from Morocco, Algeria, and Tunisia. He listed A. aff discors from Madrid based on two specimens (Warncke 1976), but later discarded these records, giving only two points in southern Spain in his distribution maps in addition to the points from north-western Africa (Gusenleitner and Schwarz 2002). I have examined 14 female and six male specimens of A. discors, all from southern parts of Portugal and Spain (Algarve, Alto Alentejo, Cádiz, Huelva, Málaga, Sevilla), where it is active from late February to April.

The situation is confused due to the status of two additional taxa described by Pérez, A. (Melandrena) creberrima Pérez, 1895 and A. (Melandrena) vachali Pérez, 1895. Andrena creberrima was described from Algeria like A. discors. Warncke (1967) designated a lectotype for A. creberrima (Fig. 12), but did not understand the concept of this species, listing only the type locality of Bône in Warncke (1974). He later gave a distribution map of Algeria (the holotype point), Italy, and Greece (Crete). Examination of the lectotype shows that the specimen is slightly paler than the type of A. discors, with fewer dark intermixed hairs on the scutum (Fig. 12B), and the tergal hair bands in better condition and therefore slightly more pronounced (Fig. 12D). This can be contrasted with the type of A. (Melandrena) bodemeyeri (Fig. 13; known synonym of A. discors, see Warncke 1967) which is comparatively dark. This kind of variation in hair colour is typical for the subgenus Melandrena, and has led to the large number of synonymous names being described, for example for A. (Melandrena) flavipes Panzer, 1799.

Grouping A. creberrima under A. discors is straightforward. The problem comes with A. vachali which was classically considered to be distinct from A. discors and with a greater range extending from the Canary Islands to southern Israel (Gusenleitner and Schwarz 2002). Moreover, type material for A. vachali has not actually been designated, as Warncke (1967) could not find material in the MNHN collection. Examination of the MNHN showed several specimens of A. vachali in the Vachal collection from Foum Tatahouine [Tataouine, Tunisia] labelled “Andrena vachali type Pérez” in the handwriting of Vachal (Fig. 14). One specimen was labelled as a lectotype with a handwritten ‘A. vachali’ determination label by H. Teunissen (Fig. 14A), but this designation was never published. Pérez (1895) did not mention the collection locality, and Warncke (1967) indicated that the species was described from Algeria. However, Gusenleitner and Schwarz (2002) give the locus typicus as Tunisia. Examination of the catalogue of Pérez did not reveal an entry for A. vachali, as is often the case, either because Pérez did not write one, or because the information is hidden under a name that differs from the published name (see below for the case of A. (Plastandrena) asperrima Pérez, 1895). It is therefore very difficult to know the type locality with certainty. However, on the basis that the MNHN specimens are labelled as ‘type’ by Vachal, they were present in his collection, and the species was named after Vachal, the position is taken that these represent genuine syntypes. The female specimen labelled as ‘type’ by Vachal is chosen as a lectotype, by present designation. Morphologically, it corresponds to the concept of A. vachali, with black pubescence on the face and mesepisternum, with reddish-brown pubescence on the remaining parts of the mesosoma, and with predominantly reddish-brown tibial scopae, though this specimen shows some darker hairs dorsally (c.f. Fig. 13D); specimens of A. vachali can be found with entirely orange-red tibial scopae. Comparison of sequences from A. discors from southern Portugal and two A. vachali from southern Israel show a very small difference of 0.91% (range 0.91–0.91%; Fig. 10). These three specimens group together and have posterior support of 100, and are clearly separated from the three other former Zonandrena species that group together in this clade, namely A. flavipes, A. (Melandrena) gravida Imhoff, 1832, and A. (Melandrena) vulcana Dours, 1873.

Taking a step back, the differentiation between A. creberrima, A. discors, and A. vachali has almost always been based on colouration, with A. discors the darkest, A. vachali the lightest, and A. creberrima somewhat intermediate. Males are generally rarer in collections; taken together, I have examined 215 female specimens, but only 73 male specimens of this discors-creberrima-vachali group. However, examination of males has led me to the conclusion that there are no apparent differences in the males of A. creberrima, A. discors, and A. vachali, and indeed the male of only one of these nominal taxa was actually originally described. All have white hairs over the majority of the face with clear black hairs laterally along the inner margins of the compound eyes, a genital capsule that is typical of the former Zonandrena with the dorsal surface of the gonocoxae with granular shagreen, with the penis valves moderately broad, and without an emargination in the outer margins of the gonostyli (Fig. 15). Indeed, the genital capsule of a syntype of A. creberrima (Fig. 15B), the barcoded A. discors from southern Portugal [IBIHM517-21] (Fig. 15C), and A. vachali specimens from Morocco have identical genital capsules (Fig. 15D). Moreover, numerous ‘aberrant’ female individuals can be found across this range with more or less pronounced tergal hair bands and variably dark or light pubescence, particularly in Crete where individuals corresponding in different ways to the concepts of A. creberrima, A. discors, and A. vachali can be found.

Based on this colour variation, the lack of variation in the male genital capsule, the unclear and overlapping distributions given by Warncke (A. creberrima and A. vachali both reported from Crete, A. creberrima and A. discors both reported from Algeria), and the very low genetic distance between Portuguese and Israeli specimens, both A. creberrima syn. nov. and A. vachali syn. nov. are synonymised with A. discors. This resolves this long-running confusion as to the correct name and identity of this taxon (Baldock et al. 2018; Wood et al. 2020a).

Distribution. Portugal, Spain (Canary Islands, mainland), Morocco, Algeria, Tunisia, Italy, Malta, Libya, Greece (Crete), Egypt, Israel, Jordan.

Material examined (illustrative). Algeria: label information illegible, possibly ‘Bone’ [= Annaba], 1♀, ZMHB (holotype/syntype of A. discors); Bône [=Annaba, 36.9092°N, 7.7264°E], 1♀, MNHN (lectotype of A. creberrima); Alger, Birmandreis [Bir Mourad Raïs], 24.iii.-30.iv.1910, 1♂, 1♀, leg. J. Bequaert, ZMHB; Greece: Krete, Anidri b. Paleochora, 200 m, 31.iii.2002, 1♀, leg. A. Müller, AMC; Krete, Palaeochora, 0–50 m, 29.iii.2002, 5♀, leg. A. Müller, AMC; Kreta, Kavalos [Kavallos], 21.v.1986, 1♀, leg. Brande, OÖLM; Crete, Dept. Heraklion, Moulia, S. of Aghia Varvara, 26.iv.1973, 1♀, leg. v. Ooststroom, RMNH; Italy: Lazio, Roma, Via Falcognana, 8.iv.1990, 1♀, leg. G.G.M. Schulten, RMNH; Roma, Via Falcognana, 15.iv.1993, 1♀, leg. G.G.M. Schulten, RMNH; Israel: Negev, 15 km S of Be’er Sheva, 31.iii.1988, 2♀, leg. R. Leys, RMNH; Jordan: Petra, 800 m, 28.ii.-4.iii.1986, 1♂, 2♀, leg. K.M. Guichard, NHMUK; Malta: Buskett [woodlands, Ħad-Dingli], 7.iii.2018, 1♀, M. Balzan Colln.; Morocco: Drâa-Tafilalet, Tazenakht, Anezal, 5 km NE Ait Igga, 21.iv.2022, 1♂, 4♀, leg. T.J. Wood, TJWC; Souss-Massa, R105, Tizirt, 8 km N, Ighir Ifran env., 12.iii.2022, 1♂, 1♀, leg. T.J. Wood, TJWC; Portugal: Foros de Vale Figueira, Montemor-o-Novo, 20.iii.2019, 1♂, leg. T.J. Wood, TJWC; Montemor-o-novo, Foros de Vale de Figuera, 24.iv.2011, 1♀, leg. A. Albernaz-Valente, TJWC; Spain: Canary Islands, Lanzarote, Haria, 6.ii.1979, 1♂, leg. Ellis & Simon-Thomas, RMNH; Bollullos Par del Condado (Huelva), 160 m, 28.iii.2009, 1♂, leg. F.J. Ortiz-Sánchez, FJOS; Puerto de Gáliz, P.N. Alcornocales (Cádiz), 425 m, 28.iv.2007, 1♀, leg. F.J. Ortiz-Sánchez, FJOS; Cádiz, Zahara de los Atunes, 12.iii.1977, 1♀, leg. Madey, NHMUK; Tunisia: Foum Tatahouine [32.9361°N, 10.4458°E], 1♀, MNHN (lectotype of A. vachali, by present designation); Oasis Gafsa [34.4350°N, 8.7678°E], 1928, 1♀, leg. B. v. Bodemeyer, MNHN (holotype of A. bodemeyeri); Fritissa, 17–19.iv.1983, 1♀, TJWC.

The treatment of A. obsoleta has a long and confused history that has caused many problems. Warncke (1967) designated a female lectotype from Babor in northern Algeria (Fig. 16). However, this lectotype designation is incorrect. In the catalogue of Pérez, under entry 1506 (p. 214), Pérez writes: “Sicile, ♀ - M’pellier ♀. – Tebessa, avril, ♀. – Biskra, mars, 3♀, rares. – Bône, ♀. – Marseille, ♂. - Bône, 27/iv ♂♀ in cop. Souci et autres [illegible – presumably a type of flower]”. Babor is not mentioned, and so Warncke’s lectotype designation is rejected. Examination of specimens labelled as obsoleta by Pérez from the localities indicated shows that they belong to a different taxon to that of Warncke’s false lectotype. Warncke’s false lectotype is a female Distandrena (=Micrandrena) with a clearly striate clypeus, and T1 is polished. This conforms to the concept of A. (Micrandrena) nitidula Pérez, 1903, and indeed Warncke (1967) synonymised A. nitidula with A. obsoleta. Only female specimens of A. obsoleta from Bône [=Annaba] in northern Algeria could be found in the Pérez collection, and almost all were in very poor condition, missing their metasomas. A single female from Bône was however in good condition, and it is here designated as a new lectotype (Fig. 17). This new lectotype specimen (i.e. the true A. obsoleta) morphologically falls within the A. (Micrandrena) mariana Warncke, 1968 complex within the former Distandrena, as it lacks striations on the clypeus. The larger A. (Micrandrena) merimna Saunders, 1908 can be excluded as this species has a very long A3 that clearly exceeds the length of A4+5, whereas members of the A. mariana complex have A3 at most slightly exceeding A4+5, A4 and A5 are sub-quadratic and slightly shorter than wide, A6–12 are as long as wide.

Warncke described A. mariana s. str. from the island of Fuerteventura in the Canary Islands, stating that the species could potentially be found in Morocco (Warncke 1968b). He then described five subspecies in subsequent publications (Warncke 1974, 1975a): A. m. mica Warncke, 1974, A. m. leptura Warncke, 1974, A. m. solda Warncke, 1974, A. m. alma Warncke, 1975, and A. m. tenostra Warncke, 1975. For North African taxa, Warncke (1974) described A. mariana solda from Morocco (Tangier) and Algeria (Algiers, including Husseyn-Dey [Hussein Dey] the locus typicus, Constantine, Bone, Blida) and Tunisia (Carthago), and noted that this species corresponds to the Andrena forms from Bone identified as iota, kappa, and lambda by Saunders (1908), though this is clearly wrong for lambda as the genital capsule illustrated by Saunders shows A. (Micrandrena) djelfensis Pérez, 1895. The A. mariana solda taxon differs from A. mariana s. str. in the female sex by the darker tergal margins, the darker antennae, and the more extensively shagreened clypeus. Warncke (1974) described A. mariana mica from southern Algeria (locus typicus Ghardaia) and Morocco, giving it a more arid distribution. It also differs from A. mariana s. str. in the female sex by the darker tergal margins and darker antennae, but the clypeus is more extensively shiny than in A. mariana solda, and the nervulus is said to be more antefurcal. Andrena m. leptura is a more eastern taxon, being described from Libya and Egypt, and having a broader process of the labrum than any of these species.

Examination of A. mariana solda material reveals that it is conspecific with the newly designated lectotype of A. obsoleta and is synonymised with it syn. nov. As identified by Warncke, the taxon has a wide distribution across Mediterranean parts of Morocco, Algeria, and Tunisia (Warncke 1974, as A. m. solda). Pérez’s reference to specimens from France indicates that his original syntypic series would have been polytypic, or he may have changed his mind between writing this catalogue entry and his 1895 publication, or even after this point. Warncke (1974: 40) indicated that A. m. solda is present in Sicily which is supported by a newly examined specimen (see below), though he did not designate any Italian specimens as paratypes. As no specimens from France labelled as obsoleta could be found in Pérez collection, the conservative position is taken here that this species does not occur in France. The implications for Warncke’s misinterpretation of A. obsoleta are discussed below.

As A. mariana solda is actually a synonym of a valid species that was described prior to A. (Micrandrena) mariana s. str., this has implications for the species-status of the other taxa lumped under A. mariana by Warncke. Genetic sequences were available for A. mariana mica Warncke, 1974 from southern Morocco, A. mariana s. str. from south-western Morocco, and A. mariana alma Warncke, 1975 from Portugal (locus typicus near Córdoba, southern Spain; Fig. 18). Contrary to the position of Warncke (1968b), examination of material from Morocco, Algeria, and Tunisia shows that A. mariana is actually widely distributed across this region, predominantly in southern, more arid areas. Genetically, A. mariana s. str. was more closely related to A. (Micrandrena) abjecta Pérez, 1895, separated by an average genetic distance of 5.55% (range 5.49–5.61%), and was strongly separated from A. m. mica by 12.73% (range 12.50–12.95%) and A. m. alma by 10.94% (range 10.65–11.39%). Andrena abjecta + A. mariana s. str. therefore formed a clade with bootstrap support of 99. Andrena m. mica and A. m. alma formed a clade, but were strongly separated from each other by 11.26% (range 10.97–11.41%). The conclusion therefore is that each of these taxa is distinct, and so A. mica stat. nov. and A. alma stat. nov. are raised to species status. Morphologically, all three taxa can clearly be separated by the structure of the female clypeus, the colour of the antennae, the extent of lightening on the tergal margins, the strength of the scutal punctures, and the shape of the process of the labrum.

Warncke described two further subspecies of A. mariana: A. mariana leptura from Egypt and A. mariana tenostra from south-eastern Spain, the latter specifically from a single female specimen from Villajoyosa in the province of Alicante, then listing additional specimens from Almería and Murcia (Warncke 1976). The status of A. mariana leptura will be dealt with in a future publication, as its exact species concept and its eastern and western limits are unclear, though it will be a valid species as it clearly differs from A. mariana s. str.; its relationship with A. mariana s.l. specimens from southern Israel must also be clarified (Pisanty, in litt.). In Iberia, A. m. tenostra is very poorly known. The distinguishing features given by Warncke (1975a) are that the process of the labrum is narrowly trapezoidal and that the clypeus is arched and somewhat flattened in the middle, in contrast to A. alma where the clypeus is largely flattened. Additional female specimens were found in Granada which highlight a difference not noted by Warncke, which is that the scutellum is polished and shiny, whereas it is dull in A. alma, and more broadly the scutal punctures are less strongly pronounced. Given that subtle differences within this group represent large genetic differentiation, A. tenostra stat. nov. is raised to species status.

Finally, a number of specimens were found in the very south of Spain (Málaga, Sevilla) which show a morphology very close to that of A. mica. However, ecologically this does not make sense as A. mica is not known from the more humid and Mediterranean areas north of the Atlas Mountains. Examination of these specimens shows that A3 is much longer than A4+5, whereas it is as long as A4+5 in A. mica. These specimens probably represent an additional undescribed species in the A. mariana complex, but they are not described at this time, as it would be beneficial to have barcoded specimens to confirm this differentiation. They are therefore referred to as aff mica in the identification key.

It is important to note that this material does not belong to A. abjecta, the status of which in Iberia is unclear. Warncke (1976) did not record this species from Iberia, and only his distribution maps (Gusenleitner and Schwarz 2002) indicate the presence of this species in northern and eastern Spain. This distribution does not make sense given that all other known records of A. abjecta come from north-western Africa (Morocco, Algeria, Tunisia), so the species would naturally be expected to occur in southern Iberia. Dardón (2014) listed two female specimens from Huelva (22.iv.1987) and Zaragoza (9.iii.1991). The specimen from Huelva may represent the undescribed aff mica taxon, but it is more difficult to conclude on the specimen from Zaragoza. For now, A. abjecta is retained on the Iberian list, but I have not examined any specimens and I consider its presence in Iberia doubtful.

Material examined. Andrena alma: Portugal: Albandeira, near Lagoa, 20.iv.2005, 1♀, leg. D.W. Baldock, TJWC; Algarve, Casaqueimada (7 km N of Silves), 20.iii.1995, 1♀, leg. T. & M. Simon Thomas, RMNH; Algarve, Monte Gordo, Retur, Praia do Cabeço, 29.iii.2022, 1♂, leg. T.J. Wood, TJWC; Algarve, Tavira, Cacela Velha, 28.iii.2022, 1♂, leg. T.J. Wood, TJWC; Castelo Branco, Fundão Vale de Prazeres, 17.iv.2021, 1♀, leg. C. Siopa, FLOW; Spain: El Soldado, Córdoba [38.3747°N, -5.0001°E], 1♀, leg. Seyrig, OÖLM (holotype); Almodóvar del Campo (Ciudad Real), 700 m, 24.iii.2005, 1♂, leg. F.J. Ortiz-Sánchez, FJOS; Santa Ana la Real, Sierra Aracena (Huelva), 630 m, 13.iv.2006, 2♂, leg. F.J. Ortiz-Sánchez, FJOS; El Hongo (P.N. Doñana), 30.iii.2018, 1♂, leg. F. Molina, EBDC.

Andrena mariana s. str.: Algeria: Saida, 15 km S of Sfissifa, Ben Ikhou, st. 6, 6.iv.1983, 6♀, leg. R. Leys & P. v. d. Hurk, RMNH; 23 km NE of Ras El Ma, st. 38, 26.iv.1983, 1♀, leg. R. Leys & P. v. d. Hurk, RMNH; Dayet el Kerch [Daïet el Kerch], st. 5, 5.iv.1983, 2♀, leg. R. Leys & P. v. d. Hurk, RMNH; Morocco: Guelmim-Oued Noun, Guelmim, Asrir, 1 km W, 20.iii.2022, 1♀, leg. T.J. Wood, TJWC; Spain: Fuerteventura, Valley Granadillos [28.3864°N, -14.0865°W], 17.iv.1934, 1♀, OÖLM (holotype); Tunisia: Ksar Hadada, 4–5.iv.1996, 46♀, leg. K. Deneš, OÖLM; M’saken, 20–21.iv.1998, 2♀, leg. K. Deneš, OÖLM; Wadi Ram, 45 km E Douz, 4.iv.2006, 1♀, leg. J. Straka, OÖLM.

Andrena mica: Algeria: Ghardaia [32.5047°N, 3.6419°E], 1♀, OÖLM (holotype); Morocco: 30 km E Midelt, 13.v.1995, 1♀, leg. Mi. Halada, OÖLM; Drâa-Tafilalet, Ouarzazate, P1506, Telouet, Adaha, 1700 m, 18.iv.2022, 6♀, leg. T.J. Wood, TJWC; Drâa-Tafilalet, Ouarzazate, 2 km W Agouim, 1800 m, 13.iv.2022, 3♀, leg. T.J. Wood, TJWC; Ifkern, 25 km E Boulemane, 24.v.1995, 1♀, leg. Mi. Halada, OÖLM; M’rirt (30 km N), 11.iii.1989, 1♀, leg. H. Teunissen, RMNH; Tunisia: Hammamet env, 15.iii.1996, 1♀, leg. K. Deneš, OÖLM; Kasserine, 13.iv.1998, 1♀, leg. K. Deneš, OÖLM; Nefta [Naftah], 20.v.1993, 1♀, leg. J. Batelka, OÖLM.

Andrena aff mica: Spain: 40 km W Málaga, Yunquera, 800 m, 29.iv.2003, 5♀, leg. J. Halada, OÖLM/TJWC; La Corchuela (Dos Hermanas, Sevilla), 35 m, 27.iii.2009, 2♂, 2♀, leg. F.J. Ortiz-Sánchez, FJOS; Río Blanco, Aguadulce (Sevilla), 300 m, 17.v.2008, 1♀, leg. F.J. Ortiz-Sánchez, FJOS.

Andrena nitidula: Algeria: Babor, 1♀, MNHN (false lectotype of A. obsoleta).

Andrena obsoleta: Algeria: Bône [=Annaba, 36.9092°N, 7.7264°E], 1♀, MNHN (lectotype of A. obsoleta, by present designation); 23 km NE of Ras El Ma, st. 38, 26.iv.1983, 1♀, leg. R. Leys & P. v.d. Hurk, TJWC; 5 km N of Mecheria, Djebel Antar, st. 10, 7.iv.1983, 1♀, leg. R. Leys & P. v.d. Hurk, RMNH; Italy: Sicilia orient., Taormina, 200 m, Ghardino [Giardini Naxos], 25.iii.1950, 1♂, Hartig & Grisham, MZUR; Tunisia: Tunis, 1911, 3♀, leg. O. Schmiedeknecht, SMFD.

Andrena tenostra: Spain: Villajoyosa [38.5097°N, -0.2299°E], 11.v.1936, 1♀, leg. Andréu, OÖLM (holotype); Salobreña, Granada, 8.v.1983, 3♀, leg. W. Perrandin, OÖLM/TJWC.

The status of these two species has been extensively argued over, and despite much attention the position remains somewhat unclear. Warncke (1967) argued that A. distinguenda Schenck, 1871 was an unavailable name because it was preoccupied by A. distinguenda Schenck, 1853. He then authored the replacement name A. obsoleta spongiosa Warncke, 1967, using the incorrect concept that A. obsoleta was the oldest available name for this group of species, as discussed above. Schönitzer et al. (1992) clarified the situation, and argued that A. distinguenda Schenck, 1853 was a nomen nudum because it appears without a description in a discussion concerning species related to A. bicolor. As a nomen nudum, the name is available and cannot preoccupy its later usage. Therefore, the name A. distinguenda Schenck, 1871 is both available and valid as the senior synonym; A. obsoleta spongiosa is consequently an unnecessary replacement name.

Andrena distinguenda was described from western Germany, and A. nitidula was described from south-western France (lectotype from Bordeaux [though this is not indicated on the specimen, it bears the number ‘675’ which refers to the entry for A. nitidula in the catalogue of Pérez], designated by Warncke 1967, Fig. 19). As discussed above, Warncke (1967) incorrectly recognised a specimen that morphologically corresponds to A. nitidula as the lectotype of A. obsoleta, and then synonymised A. nitidula with A. obsoleta. He then treated A. distinguenda Schenck, 1871 as a subspecies of A. obsoleta in the form A. obsoleta spongiosa. He then later adopted a three subspecies model, A. o. obsoleta (North Africa), A. o. nitidula (Western Europe), and A. o. spongiosa (Central Europe) (see Schönitzer et al. 1992), finally introducing a fourth subspecies A. o. puella Alfken, 1938 in Sardinia and Sicily (see Gusenleitner and Schwarz 2002). Schönitzer et al. (1992) argue that both A. distinguenda and A. nitidula can be morphologically separated in north-western Italy (Bordighera). They argue that A. nitidula occurs in North Africa, citing Pérez (1903) and Alfken (1914). They also argue that specimens from southern Spain (Tarifa) appear to be somewhat intermediate between A. distinguenda and A. nitidula, concluding that the question as to whether there are two species or simply a very variable single species remains open.

Burger and Herrmann (2003) revised European material from this species pair using the names A. distinguenda and A. nitidula, after having examined syntypes of A. nitidula in the ZMHB collection. Using morphological characters, they conclude that (in a European context) A. distinguenda has a predominantly eastern distribution, from Central Europe to Greece, with populations also in the south of France and a few records from north-eastern Spain. In contrast, A. nitidula is a western species, from Iberia to the south and east of France and into south-western Germany and the north-west of Italy. They reach the same conclusion as Schönitzer et al. (1992), that where these two forms occur in sympatry (Germany, France, Italy) they can be morphologically separated. They do not conclude on the status of A. obsoleta, but note that A. nitidula occurs in North Africa, and that Warncke’s treatment is unclear. This problem arises because it does not appear that any of the authors of Schönitzer et al. (1992) or Burger and Herrmann (2003) inspected the lectotype of A. obsoleta designated by Warncke in the MNHN. This lectotype bears a handwritten label that the lectotype designation is incorrect, and that the specimen corresponds to A. distinguenda Schenck (Fig. 16A). However, it is unclear who wrote this, as neither Schönitzer et al. (1992) nor Burger and Herrmann (2003) mention this explicitly, and I do not recognise the handwriting on the label.

Barcode analysis complicates this matter further (Fig. 18). Analysis of sequences from France, Germany, Morocco, Portugal, and Spain shows the presence of two clear clades. One clade is composed of specimens identified as A. nitidula from France, Portugal, and Spain, including specimens from the south of Iberia (Algarve, Málaga, Murcia). This clade would appear to be the true A. nitidula. However, the second clade contained sequences from specimens identified as A. distinguenda from central Germany, but also from a specimen identified as A. nitidula from central Spain (Guadalajara) and two specimens identified as A. nitidula from northern Morocco (Fès-Meknes, Oriental). These two clades were consistently separated genetically, by an average genetic distance of 7.08% (range 5.58–8.77%). The nominally true A. nitidula clade showed low intraspecific variation of 0.26% (range 0.00–0.63%). However, the situation in the clade containing A. distinguenda sequences from Germany showed intraspecific variation of 2.84% (range 0.00–4.10%), with the 4.10% genetic distance recorded between specimens from Germany and the Middle Atlas (Bakrit, near Azrou) in Morocco.

Morphologically, both Spanish and Moroccan specimens falling into the A. distinguenda clade conform to the concept of A. nitidula using the criteria specified by Burger and Herrmann (2003). Indeed, all material I have identified from north-western Africa morphologically conforms to the concept of A. nitidula sensu Burger and Herrmann. Although the genetic data strongly supports the existence of two species, the conflict between the genetic and morphological results suggests that, at the moment, they cannot be consistently separated morphologically across their range. No taxonomic action is taken here, and the morphological criteria of Burger and Herrmann (2003) are followed for the purpose of the identification key. In Iberia, A. nitidula is considered to be present throughout, but A. distinguenda is limited to the extreme north-east of Spain. However, it is clear that there is a major disagreement between the morphological and genetic methods, and these two species may ultimately not be consistently separable. Further analysis using more powerful genetic techniques is necessary.

Inspection of Micrandrena specimens from high altitude in the Sierra Nevada revealed the presence of a species that morphologically resembles A. (Micrandrena) rugulosa Stöckhert, 1935 due to its head that is only slightly shorter than wide rather than clearly shorter than wide, an unusual character in Micrandrena. This finding is remarkable, because although widely distributed in Central and Eastern Europe, A. rugulosa has a western limit in the Swiss Alps, and has not been previously recorded from France, Spain, or Portugal (Gusenleitner and Schwarz 2002). Samples collected from the Sierra Nevada at high altitude (2000–2100 m) during 2021 are strongly differentiated from A. rugulosa, separated by an average genetic distance of 16.01% (range 15.81–16.41%; Fig. 18). Instead, they form a clade with bootstrap support of 96 with sequences of A. (Micrandrena) atlantea Wood, 2021 (High and Middle Atlas in Morocco) and a morphologically similar undescribed Micrandrena from high altitude (1800–1900 m) in the Moroccan Middle Atlas.

The Sierra Nevada sequences were separated from A. atlantea by an average genetic distance of 6.04% (range 5.71–6.38%) and from the undescribed Micrandrena by an average genetic distance of 6.34% (range 6.31–6.38%). These three species would therefore seem to represent an isolated Micrandrena lineage that is restricted to the Sierra Nevada and the High and Middle Atlas Mountains in Morocco, with consequent genetic and morphological divergence. This is the same pattern as observed in the subgenus Euandrena, suggesting that the Sierra Nevada hosts the remnants of a fauna that was presumably once more widespread across Iberia and North Africa. Additional sampling and genetic analysis is required to determine if this pattern holds true for other bee groups. The new Micrandrena species from the Sierra Nevada is described below, and the undescribed Micrandrena species from Morocco will be described in an upcoming publication.

Andrena niveata was described from Germany and Hungary, without further information (Friese 1887). The exact type locality is unclear, as well as the type depository, with no clearly identifiable syntypes located during searches in the SMFD and ZMHB collections, the suggested possible depositories (Gusenleitner and Schwarz 2002). Nevertheless, the concept of A. niveata in Central Europe is clear, and A. niveata is a well-defined species in this region (Schmid-Egger and Scheuchl 1997; Gusenleitner and Schwarz 2002; Amiet et al. 2010). Warncke described three subspecies; A. n. lecana Warncke, 1975 (Spain; locus typicus: Rivas-Vaciamadrid), A. n. haloga Warncke, 1980 (Italy; locus typicus: Lecce), and A. n. bubulca Warncke, 1975 (Turkey; locus typicus: Erzurum); the status of the Italian and Turkish subspecies is not addressed here. Unfortunately, all nine freshly collected A. n. lecana specimens from central and southern Spain (including from the locus typicus) sent for genetic analysis failed to produce sequences. More specific primers will be required to address this question using genetic evidence.

Warncke (1975a) described A. n. lecana extremely briefly, noting that in both sexes the first tergum was finely but clearly shagreened, and finely but more strongly punctured than in A. niveata s. str., with the following terga also more clearly and strongly punctured. He then gave a distribution of A. n. lecana across much of central Spain south to hilly parts of the provinces of Granada (Huéscar) and Almería (Tíjola). In contrast he gives a distribution for A. niveata s. str. of the whole of Iberia, though his distribution maps (Gusenleitner and Schwarz 2002) show a distribution covering eastern, central, and south-eastern Spain. There is therefore the possibility that the two subspecies exist in sympatry, suggesting that they may be distinct.

Examination of material from Spain shows that the two taxa are distinct, but morphological separation is challenging, may not be possible in all cases when old, abraded, or dirty specimens are available, and is best made with reference to the male genital capsule. Taking male specimens, the shape of the gonostyli are distinctive. In A. niveata s. str., the gonostyli are elongate, with the inner margins of the gonostyli only weakly produced towards the penis valves (Fig. 20A). In contrast, in A. n. lecana, the inner margins of the gonostyli are strongly and clearly produced towards the penis valves (Fig. 20B); the apical spatulate part of the gonostyli are also comparatively shorter, thus the gonostyli appear shorter and more compact in A. n. lecana and longer and more elongate in A. niveata s. str. The structure of the genital capsule of A. niveata s. str. is consistent across Spain to the province of Málaga (the most southerly A. niveata s. str. male examined). Furthermore, the male scutum is shagreened in A. niveata s. str., whereas it is extensively shiny in A. n. lecana, and the terga are also more strongly shagreened in A. niveata s. str. compared to A. n. lecana (Fig. 20C, D). Separation of females (recognised by their wide and poorly defined propodeal triangle, Fig. 20E, F, in combination with their depressed tergal margins) is more challenging and may not be possible in all situations. Typically, A. niveata s. str. has the terga shagreened, with punctures on the tergal discs partially disappearing into this shagreenation (Fig. 20G). In contrast, A. n. lecana has the terga shiny, almost without shagreen, with the dense punctures on the tergal discs strong and clearly visible, not disappearing due to the absence of background shagreenation (Fig. 20H). The scutum and scutellum are also more extensively shiny, whereas in A. niveata s. str. these areas usually show at least some shagreenation and are never polished and smooth. The problem comes that some individuals nominally assigned to A. niveata s. str. have partially shing terga, and without genetic sequences it is not possible to say with complete confidence if they belong to A. niveata s. str. or A. n. lecana. Moreover, all Iberian A. niveata s. str. have tergal shagreenation that is weaker than that of A. niveata s. str. populations in Central Europe.

The position is taken here that the strength of tergal shagreenation in A. niveata s. str. females is variable across Europe, but that A. lecana stat. nov. is a valid species based on the overlapping range in combination with the consistent difference in the shape of the male genital capsule. It has a distribution across steppic parts of central Spain, extending into mountainous areas in south-eastern Spain. Females displaying any level of shagreenation on the tergal discs are considered to represent Iberian populations of A. niveata s. str., and females with completely polished tergal discs without a trace of shagreenation represent A. lecana (see identification key). Future genetic investigation using more targeted primers will be necessary to confirm this position. Finally, specimens of A. lecana from high altitude in the Sierra Nevada show slightly different antennal ratios in the male sex, though the male genital capsule is otherwise identical; this requires further investigation.

Material examined. Andrena lecana: Spain: Ribas [Rivas-Vaciamadrid, 40.3503°N, -3.5390°E], 6.v.1908, 1♀, leg. Dusmet, OÖLM (holotype); Carboneras de Guadazaón (Cuenca), 1030 m, 16.v.2009, 1♀, leg. F.J. Ortiz-Sánchez, FJOS; Guadalajara, Alcolea del Pinar, 12.v.2021, 1♀, leg. T.J. Wood, TJWC; Guadalajara, Lupiana, 12.v.2021, 1♀, leg. T.J. Wood, TJWC; Huéscar (Granada), 1900, 1♂, leg. Escalera, OÖLM (paratype); La Cabrilla, Sierra Cazorla (Jaén), 1600 m, 1–3.vi.2022, 1♀, leg. C.M. Hererra, CMHC; Madrid, Rivas-Vaciamadrid, Canal de Manzanares to Camino de Uclés, 19.v.2021, 2♀, leg. T.J. Wood, TJWC; Órgiva, N, 1300 m, Sierra Nevada, 26.vi.1988, 1♀, leg. M. Schwarz, OÖLM; Pozuelo, La Fuente, 1♂, OÖLM (paratype); Segovia, Madrona, 500 m NE, Arroyo del Hocino, 15.v.2021, 2♀, leg. T.J. Wood, TJWC; Sierra Nevada, Trevélez, Refugio La Campiñuela, 2400 m, 14.vi.2021, 7♂, 2♀, leg. T.J. Wood, TJWC.

Andrena niveata s. str.: Spain: Cáceres, Cuacos de Yuste, 500 m, 11.v.1999, 1♀, leg. H. & J.E. Wiering, RMNH; Cádiz, Grazalema, Río Guadalete, 28.v.2021, 1♀, leg. T.J. Wood, TJWC; Cádiz, Embalse de Barbate Sw, 1♂, 3♀, 6.v.2017, leg. Barták & Kubik, OÖLM/TJWC; Granada, Sierra de Baza, Prados del Rey, 2000 m, 19.vi.2021, 11♀, leg. T.J. Wood, TJWC; Málaga, 5 km W of Alhaurín de la Torre, 4.vi.1962, 1♀, leg. Jeekel & Wiering, RMNH; Málaga, Cortes de la Frontera, path to Llano de las Labores, 26.v.2021, 3♀, leg. T.J. Wood, TJWC; Málaga, Estepona, 21.iv.1983, 1♀, leg. H. Teunissen, RMNH; Navarra, Tudela, 16.iv.1978, 1♀, leg. C. Gielis, RMNH; Gerona, Figueras, 15.iv.1971, 1♀, leg. J. Leclercq, UMONS.

Members of this species group are challenging to identify and have been inconsistently treated in the literature, with variable species concepts. Andrena spreta was described from Algeria, with Warncke (1967) designating a lectotype from Biskra (Fig. 21A; Pérez also mentions a female from Constantine in his catalogue). Andrena curtula was described from north-eastern Spain, with Warncke (1967) designating a lectotype from Barcelona (Fig. 21B). Andrena pusilla was described from the south of France (Bordeaux, Nantes) and Spain, with Warncke (1967) designating a lectotype from Nantes (Fig. 21C). Finally, A. pauxilla was described from the Upper Rhine valley in south-western Germany, with Stöckhert (1935) designating a female from Karlsruhe as the holotype.

Warncke (1967) recognised three species, A. spreta, A. curtula, and A. pauxilla, treating A. pusilla as a subspecies of A. spreta. Warncke (1974) recognised only one taxon in North Africa (A. spreta spreta), but Warncke (1976) surprisingly recognised only A. spreta pauxilla in Iberia, not mentioning A. curtula as a species despite its description from Iberia, only referring to its listing in Ceballos (1956) but indicating that this referred to A. (Micrandrena) tenuistriata Pérez, 1895. This treatment is difficult to understand. Warncke later adopted an extremely broad position (see distribution maps in Gusenleitner and Schwarz 2002), lumping all members of the group into a broad A. spreta. In this model, A. spreta s. str. was restricted to North Africa, A. spreta curtula was distributed across Iberia to southern France, A. spreta pusilla was found in northern and Central Europe, with two more subspecies in i) Italy and ii) south-eastern Europe to the Levant.

This position was not adopted by subsequent authors. Schwarz et al. (1996) accepted A. pusilla as a valid species, as well as A. curtula. However, they synonymised A. pauxilla with A. curtula. Schmid-Egger and Scheuchl (1997) did not follow this interpretation, treating A. pauxilla as a distinct species, though Amiet et al. (2010) followed the position of Schwarz et al. (1996). Dardón (2010) and Dardón et al. (2014) accepted only a broad A. spreta taxon, including A. curtula, A. pusilla, and A. pauxilla, recognising A. spreta s. str. and A. spreta pusilla in an Iberian context. The situation is therefore unclear, both for the number of species present in this complex, and their distributions.

Analysis of barcodes provides unambiguous support for the existence of three distinct species (Fig. 18), A. spreta including A. curtula, A. pusilla, and A. pauxilla sp. resurr. Specimens of A. spreta from Israel, Morocco, Portugal, Spain formed a clear clade with bootstrap support of 85. These sequences were separated by an average genetic distance of 0.45% (range 0.00–1.07%). They were clearly separated from a sister clade of A. pusilla sequences from Belgium, France, and Germany by an average genetic distance of 4.33% (range 3.88–4.84%). The A. pusilla clade had low intraspecific genetic distance of 0.29% (range 0.00–0.71%). Andrena pauxilla was strongly differentiated and was found as sister to A. (Micrandrena) rugulosa, showing average genetic differentiation of 10.96% (range 10.38–11.85%) from A. spreta and 11.69% (range 10.75–12.59%) from A. pusilla. This A. pauxilla clade includes a specimen from the south of France (KJ837178, Lac St. Croix) that was identified as A. curtula, the genetic results suggest that it is misidentified. Morphologically, A. pauxilla is clearly differentiated from A. spreta and A. pusilla due to the absence of a gradulus at the base of the terga. Confusion is impossible in the female sex after recognition of this character; the foveae are also longer and narrower and filled with white rather than light brown hairs, the scutum is more densely and uniformly punctate when compared to A. spreta. Examination of the lectotypes of A. spreta and A. curtula shows no clear structural difference. The degree of shagreenation varies, but this is typical for A. spreta across its range; both lectotype specimens show the presence of a gradulus at the base of the terga, excluding their conspecificity with A. pauxilla. Due to these genetic results combined with the morphology of the lectotype specimens, the synonymy of A. curtula with A. spreta as proposed by Dardón (2010) and Dardón et al. (2014) is therefore followed, though their synonymies of A. pusilla and A. pauxilla with A. spreta are rejected.

A single barcode was available from Moroccan specimens from the Middle Atlas tentatively identified as A. pauxilla which showed an average genetic distance to European A. pauxilla specimens of 2.00% (range 1.48%–2.29%). This is considered to represent only separation by distance, and thus A. pauxilla is recorded for the first time in North Africa and unambiguously recorded in Spain. Within Spain, A. pauxilla appears to be principally recorded from mountain ranges such as the Sierra de las Nieves, the Sierra Nevada (Fig. 22A, B), Sierra Cazorla, Sistema Central, and Sistema Ibérico (Fig. 22C, D). Andrena pauxilla would therefore have a currently known distribution of Morocco, Spain, France, and Germany. On the basis of these results, A. spreta would appear to be a pan-Mediterranean species, and A. pusilla its predominantly northern counterpart (though the species descends south into Italy). The two species can be found in sympatry in the south of France (Praz, in litt.), but to date I have examined no A. pusilla specimens from Iberia. Due to the confusion in this group, it must be considered absent until it can be positively demonstrated to be present south of the Pyrenees.

Material examined. Andrena curtula: Spain: Barcelona [41.4028°N, 2.1332°E], 1♀, MNHN (lectotype; Fig. 21B).

Andrena pauxilla: France: Bischenberg, 28.vi.1936, 1♀, leg. M. Klein, det. E. Stöckhert, OÖLM; Hausbergen, 29.vi.1930, 1♀, leg. M. Klein, det. H.R. Schwenninger, OÖLM; Morocco: Fès-Meknès, Boulemane, R503, 7 km SE of Boulemane, 1900 m, 22.v.2022, 1♂, 1♀, leg. T.J. Wood, TJWC; Spain: Canet de Mar, 26.iii.1963, 1♀, leg. F. Vergés, det. H.R. Schwenninger, OÖLM; Cuenca, Huerta del Marquesado, environs north of town, 26.vi.2021, 3♀, leg. T.J. Wood, TJWC; Granada, Sierra Nevada, Jardín Botánico Hoya de Pedraza environs, 1900 m, 9.vi.2021, 1♀, leg. T.J. Wood, TJWC; Granada, Sierra Nevada, Mirador Monte Ahí de Cara, 2100 m, 12.vi.2021, 1♀, leg. T.J. Wood, TJWC; Guadalajara, Aldeanueva de Atienza, 9.vii.2021, 1♀, leg. T.J. Wood, TJWC; Málaga, PN Sierra de las Nieves, mountain peak S of Pinsapo Escalereta, 30.v.2021, 1♀, leg. T.J. Wood, TJWC; La Cabrilla, Sierra Cazorla (Jaén), 1600 m, 3.vi.2022, leg. C.M. Herrera, CMHC; Sierra Cazorla, Puerto Llano, 1800 m, 11.vi.2022, 1♂, 3♀, leg. J. Valverde, CMHC; Teruel, Guadalaviar, Rambla de los Ojos, 27.vi.2021, 1♀, leg, T.J. Wood, TJWC; Teruel, Villar del Cobo, Barranco de los Oncenachos, 27.vi.2021, 6♀, leg, T.J. Wood, TJWC; Cuenca, Mirador Valle de Valdecabras, 21.vi.2021, 2♀, leg, T.J. Wood, TJWC.

Andrena pusilla: France: Nantes [47.2233°N, -1.5542°W], 1♀, MNHN (lectotype; Fig. 21C).

Andrena spreta: Algeria: Biskra [34.8600°N, 5.6995°E], 1♀, MNHN (lectotype; Fig. 21A).

Andrena strohmella was described from southern Germany and is a typically early spring species in the Central European Andrena fauna, with records extending south to the High and Maritime Alps in France, and west to the Bordeaux region; it has not previously been reported from the Pyrenees (Gusenleitner and Schwarz 2002). Warncke (1974) later described A. icterina from northern Algeria. He later indicated the presence of this species in south-eastern Spain in his distribution maps (Gusenleitner and Schwarz 2002), though he did not list its present in Iberia (Warncke 1976). Wood et al. (2020b) later reported the presence of this species in northern Morocco. In his description of A. icterina, Warncke draws comparison with A. strohmella, noting that the process of the labrum is half as broad, the clypeus is more sparsely punctate and lacking an unpunctured midline, that the terga are more strongly shagreened, and that the carinae on the dorsolateral corners of the first tergum are more weakly pronounced. Dardón (2010) and Dardón et al. (2014) did not actually examine any material of Iberian A. icterina, but inspection of new Iberian material has demonstrated that A. icterina is more widespread in Iberia than previously thought, being found from the Los Alcornocales to the Sierra de las Nieves, the Sierra Nevada, Sierra de Baza, Sierra Cazorla, and Sistema Central, and is also newly reported from northern Portugal (see below). In all of these localities it is found in mountainous areas, generally at high altitude in excess of 1000 m.

Dardón (2010) and Dardón et al. (2014) reported A. strohmella from north-eastern Spain, from Tarragona: Alcanar, v.2002, leg. Kadlec, OÖLM. I have not been able to locate and examine this specimen for myself in the OÖLM collection, but it is slightly problematic for ecological regions. Andrena strohmella is clearly a species of deciduous temperate woodland; this is reflected in its distribution across Central Europe, and in its association with flowering trees such as Crataegus (Rosaceae), Prunus (Rosaceae), and Salix (Salicaceae). Andrena strohmella has therefore not previously been reported from Mediterranean regions. The record from Alcanar does not therefore match the ecology of this species, Alcanar being a low-elevation (c. 70 m) village close to the Mediterranean coast. However, A. strohmella is present in north-eastern Catalonia based on recently collected specimens (Álvarez Fidalgo, in litt.), and so the species is retained on the Spanish and Iberian lists.

More broadly, morphological differences between A. strohmella and A. icterina are slight, though they do not appear to introgress based on examined specimens. Some characters such as the strength of the carinae on the dorsolateral corners of the first tergum are not completely consistent, because some specimens in Central Europe can be found in which these are very weakly produced. Genetic data will hopefully clarify the status of A. icterina, but unfortunately, like A. lecana, this taxon appears to be challenging to barcode, as all seven Iberian specimens sent for genetic analysis failed or returned corrupted sequences.

Material examined. Andrena icterina: Portugal: Bragança, Serapicos, 16.v.2021, 1♀, leg. A. Soares, A. Soares Coll.; Spain: Ávila, Hoyocasero, El Pinar de Hoyocasero, 16.v.2021, 1♀, leg. T.J. Wood, TJWC; Ávila, Navalsauz, 1 km E, Alberche stream, 16.v.2021, 1♀, leg. T.J. Wood, TJWC; Cádiz prov., Vent. L. Canillas Hozgarganta­Tal b. Jimena 250 m, 14.iv.1985, 3♀, leg. W. Schacht, OÖLM; Campamento Alfaguara (Alfacar, Granada), 1420 m, 13.v.2007, 1♂, 2♀, leg. F.J. Ortiz-Sánchez, FJOS; Cortijo Tortas, Paterna del Madera (Albacete), 1310 m, 30.iv.2022, 2♂, leg. F.J. Ortiz-Sánchez, FJOS; Granada, Sierra de Baza, Prados del Rey, 2000 m, 19.vi.2021, 1♀, leg. T.J. Wood, TJWC; Granada, Sierra Nevada, Capileira to La Cebadilla, 1500 m, 8.vi.2021, 1♀, leg. T.J. Wood, TJWC; Granada, Sierra Nevada, Jardín Botánico Hoya de Pedraza environs, 1900 m, 2♀, leg. T.J. Wood, TJWC; Granada, Venta de los Alazores, 25.v.1982, 1♀, leg. R. Leys, RMNH; La Cabrilla, Sierra Cazorla (Jaén), 1600 m, 3.vi.2022, leg. C.M. Hererra, CMHC; Málaga, PN Sierra de las Nieves, mountain peak S of Pinsapo Escalereta, 1600 m, 30.v.2021, 3♀, leg. T.J. Wood, TJWC; Sierra Cazorla, Puerto Llano, 1800 m, 11.vi.2022, 1♀, leg. J. Valverde, CMHC; Puerto Crucetillas, Riópar (Albacete), 1340 m, 30.iv.2022, 1♂, leg. F.J. Ortiz-Sánchez, FJOS; Sierra de Cazorla, Nava de las Correhuelas, 1.v.2021, 1♂, leg. C.M. Hererra, CMHC.

Ortiz-Sánchez et al. (2022) reported A. nigrospina as new for Spain from the Sierra de Cazorla based on male specimens with their distinctive genital capsule. Wood (2023a) further reported two barcoded specimens from Spain (Sistema Central and Sistema Ibérico) belonging to A. nigrospina. This genetic result was further supported with the addition of more sequences from Morocco (Fig. 9), revealing that A. nigrospina is also present in the Middle Atlas. The A. nigrospina clade maintained a low intraspecific genetic distance of 0.41% (range 0.00–1.06%) from Morocco to Kyrgyzstan, and average separation from A. pilipes was 1.63% (range 1.06–2.35%). As this difference is still small in absolute terms, bootstrap support was moderate, with 77 for A. pilipes and 84 for A. nigrospina, but both are considered to be distinct species. Females cannot currently be consistently separated morphologically, so no characters are given in the identification key. In Iberia, A. nigrospina is newly reported for northern Portugal from close to the Peneda-Gerês National Park. The species is likely to be restricted to cooler parts of northern, central, and eastern Iberia where it will probably be found only in mountainous regions, or at least at elevation. More sampling and genetic study is required to define its exact range limits.

Material examined. Andrena nigrospina: Portugal: Minho, Ruivães, N103, 12.v.2019, 1♂, 1♀, leg. Wood, TJWC; Spain: Cuenca, Pajaroncillo, 3 km SW, Arroyo de Peña Quebrada, 26.vi.2021, 4♀, leg. T.J. Wood, TJWC (barcoded); Guadalajara, Bustares, 2 km N, Alto Rey, 1780 m, 1♀, 9.vii.2021, leg. T.J. Wood, TJWC (barcoded).

Andrena agilissima is a widespread West Palaearctic species that is well-known in Central and Southern Europe to north-western Africa. In contrast, A. asperrima is much less well known, having a more Mediterranean distribution in France, Spain, Morocco, Algeria, and Tunisia. Unlike A. agilissima, A. asperrima is bivoltine and is exceptionally variable in the density, size, and strength of the integumental punctation. In the female sex, the typical form has strong and dense punctures on the terga, allowing easy separation from A. agilissima in which the terga have small and subtle punctures. However, many specimens of A. asperrima can be found which have greatly reduced tergal punctation and which are therefore extremely similar to A. agilissima; they can be separated by the smaller body size and the sparser punctation of the scutum. This sparsely punctate form is more common in the south-west and was described from Morocco by Warncke as A. asperrima alascana Warncke, 1974. Overall, the two species are clearly separable by their genital capsules in the male sex.

Because of this variation, it is important to ensure that Iberian material is conspecific with North African material, since the oldest names of Pérez (1895) were described based on North African material. Warncke (1967) designated a male lectotype for A. asperrima using a specimen from Biskra from the collection of Pic (Fig. 24A, B). This is unjustified, as it is not part of the original syntypic series, and Warncke’s designation is here rejected. In Pérez’s catalogue, under entry ‘1030 Andrena trachodes J.P.’ [an unpublished name; Pérez sometimes changed his mind and therefore a different name to the published name can be present in the catalogue] Pérez writes that the species comes from Constantine [in Algeria], drawing comparison with A. flessae Panzer, 1805 (= A. agilissima (Scopoli, 1770)) and arguing it differs by the stronger punctation of the metasoma. Inspection of the Pérez collection shows the presence of a female specimen from Constantine which bears the label ‘1030’, the code used by Pérez for this species. This specimen is designated as a new lectotype (Fig. 24C–F); it conforms to the concept of A. asperrima used by subsequent authors.

Andrena atricapilla Pérez, 1895 was also described from Algeria, but only in the male sex. Warncke (1967) listed this taxon as a synonym of A. asperrima, though he did not mention a lectotype. A male specimen labelled as atricapilla J.P. in the handwriting of Pérez is separated in the MNHN collection and labelled as a lectotype, probably by Teunissen as one of his determination labels is present. This lectotype designation was never published, and so it is here designated as a lectotype (Fig. 25). The synonymy with A. asperrima is maintained, as the genital capsule is typical for the species concept.

Genetic analysis of specimens of A. agilissima and A. asperrima from Croatia, Morocco, Portugal, and Spain showed two clear clades (Fig. 9). One clade contained only specimens determined as A. agilissima from Croatia, Portugal, and Spain. These sequences showed average intraspecific variation of 0.07% (range 0.00–0.25%), and were strongly separated from the second clade by an average genetic distance of 9.33% (range 8.74–10.20%). This second clade contained specimens identified as A. asperrima from Spain and Morocco, but also three specimens identified as A. agilissima from the Middle Atlas. There can be no doubt as to the identity of these specimens, as two are males, and their genital capsule is that of A. agilissima and not A. asperrima. However, the genetic differentiation within this clade is very low at an average of 0.42% (range 0.00–0.91%). As for A. bimaculata and A. tibialis, there is clearly genetic complexity here, as two species with distinct genital capsules have barcodes that form intermixed clades. No taxonomic action is taken on the basis of these barcoding results.

However, what can be concluded is that Iberian material of A. asperrima is conspecific with North African material, including the weakly punctate form that dominates in Morocco. In combination with the new lectotype designations, the invariant genital capsule, and these genetic results, the concept of Warncke (1967; 1974) and Gusenleitner and Schwarz (2002) regarding A. asperrima is maintained. The full synonymy is given below.

Andrena (Plastandrena) asperrima Pérez, 1895: 33, ♀♂ [Algeria, lectotype by present designation: MNHN].

Andrena (Plastandrena) atricapilla Pérez, 1895: 33, ♂ [Algeria, lectotype by present designation: MNHN].

Andrena (Plastandrena) flessae var. elcheensis Friese, 1922: 211, ♀ [Spain: ZMHB, not examined].

Andrena (Plastandrena) hemicyanea Cockerell, 1930: 112, ♀ [Tunisia: type lost?]

Andrena (Plastandrena) asperrima alascana Warncke, 1974: 36, ♀♂ [Morocco: OÖLM, examined].

Distribution. Spain, France, Morocco, Algeria, Tunisia.

Material examined. Algeria: Constantine [36.3645°N, 6.6409°E], 1♀, MNHN (lectotype of A. asperrima, by present designation; Fig. 24C–F); Biskra, v.1885, 1♂, leg. Blause, MNHN (false lectotype of A. asperrima; Fig. 24A, B); Biskra [34.8600°N, 5.6995°E], 1♂, MNHN (lectotype of A. atricapilla, by present designation; Fig. 25); Morocco: Tizi-n-Talrhemt nr. Midelt [32.6821°N, -2.9344°E], 1900 m, 1.vi.1968, leg. M.A. Lieftinck, OÖLM (holotype of A. asperrima alascana).

Andrena (Simandrena) cilissaeformis Pérez, 1895: 42, ♀ [Spain, lectotype by present designation: MNHN]

Andrena breviscopa auctorum.

Remarks. Andrena breviscopa Pérez, 1895 was described in the female and male sexes from North Africa. Warncke’s treatment of A. breviscopa is curious, because he designated a lectotype (Fig. 26) and listed the taxon as a synonym of A. numida Lepeletier, 1841 (Warncke 1967). However, just a few years later he listed A. breviscopa as a valid taxon in the subgenus Simandrena (Warncke 1974). Examination of the female lectotype designated by Warncke shows that his original synonymy was correct; the taxon is clearly not a Simandrena, and is indeed a synonym of A. numida.

The use of the name A. breviscopa to apply to the taxon present in Spain, Morocco, and Algeria is therefore incorrect. The correct name is A. cilissaeformis Pérez, 1895 sp. resurr. Andrena cilissaeformis was described from Spain, not Algeria as stated in Warncke (1967) and Gusenleitner and Schwarz (2002). This is because Pérez (1895) does not directly state the type locality (or indeed, any information about the collecting locality of any of the species described in this work), but this information is included in his personal catalogue. Warncke (1967) did not examine material of A. cilissaeformis, stating that whilst the label was present, material was missing. Examination of material in the MNHN has located a specimen labelled with ‘Esp’ [Espagne = Spain] in Pérez’s handwriting (Fig. 27). This specimen was separated by Teunissen, but he never published this information. This specimen is badly damaged, but it is a Simandrena and conforms to A. breviscopa auctorum sensu Warncke (1974) and subsequent publications. It is designated as a lectotype to fix the name on the Iberian population. Andrena breviscopa is returned to its original synonymy with A. numida.

Distribution. Spain, Morocco, Algeria.

Material examined. Andrena breviscopa: Algeria: Ghardaia [32.5047°N, 3.6419°E], 1♀, MNHN (lectotype; Fig. 26); (Andrena cilissaeformis): Spain: no collection information, 1♀, MNHN (lectotype by present designation; Fig. 27).

Two further issues require discussion. The first is the identity of A. poupillieri. Praz et al. (2022) identified two potential mitochondrial lineages that could correspond to this species. Additional sampling in Morocco has clarified the situation; poupillieri 1 sensu Praz et al. (2022) corresponds to A. (Taeniandrena) gregaria Warncke, 1974 and poupillieri 2 sensu Praz et al. (2022) corresponds to the concept of A. poupillieri used by Warncke. As the type of A. poupillieri is lost, it is beneficial to designate a neotype in order to fix the concept of this species in line with the existing literature; a specimen from Algeria is chosen, as this is the original locus typicus. As A. gregaria does not occur in Iberia, it is not included in the phylogenetic tree presented here; it will be dealt with in the upcoming North African Andrena revision. The species A. poupillieri is commonly encountered in Morocco, and is clearly identifiable from barcodes, forming a clade with bootstrap support of 100. Separation of females from A. (Taeniandrena) ovatula (Kirby, 1802) can be made by the tergal punctation, this being much stronger in A. ovatula. Generally, the two taxa are well-separated by an average genetic distance of 6.12% (range 5.86–7.79%). Males of A. poupillieri can be recognised based on the genital capsule, in which the gonostyli are apically produced into acute points.

Warncke changed his mind about the status of A. poupillieri – in Warncke (1967), he lists the taxon as A. ovatula poupillieri, but then in Warncke (1975a) he described A. poupillieri incana from the Balearic islands, listing A. poupillieri s. str. from southern Iberia (Warncke 1976). In his distribution maps (Gusenleitner and Schwarz 2002), he gives a distribution of southern Iberia, the Balearic islands, north-western Africa, Libya, and Crete. Records from Crete are likely to refer to unrecognised A. (Taeniandrena) ovata Schenck, 1853 (Wood, unpublished data). Examination of male specimens from the extreme south of Spain shows that A. poupillieri is present based on the distinctive genital capsule, though as females cannot be separated morphologically from A. ovatula the exact range of the two species is unclear. Andrena ovatula reaches at least as far south at the Sierra de las Nieves in Málaga province [IBIHM1045-22], though this far south it may be restricted to mountainous areas whereas the two examined male A. poupillieri specimens come from the coast. More collection is required. For now, A. poupillieri is considered to be a rare and little-collected taxon in Iberia, probably restricted to the coast in the south and south-east of the peninsula. A single sequence for A. p. incana was available from Mallorca that unambiguously nests within the North African A. poupillieri sequences; it is identical to 12 of the North African sequences, and differs from the thirteenth by just 0.15%. As such, A. poupillieri including A. p. incana shows extremely low intraspecific variation of 0.03% (range 0.00–0.15%), and subspecific status is unnecessary for the population on the Balearic islands.

Material examined. Spain: Málaga, Estepona, 24.iii.1986, 1♂, leg. J. van Oosterhout, RMNH; Málaga, San Pedro de Alcántara, 15.iii.1986, 1♂, leg. C. v. Achterberg, RMNH.

Praz et al. (2022) synonymised A. (Taeniandrena) similis Smith, 1849 with A. russula, taking a broad, pan-Mediterranean approach. In an Iberian context, sequences from central and northern Iberia clearly fall into a broad A. russula clade along with sequences from Morocco to Cyprus and Israel and north to the United Kingdom and Finland (Fig. 28). However, in the extreme south-west of Iberia in southern Portugal, sequences from specimens that are morphologically indistinguishable from A. russula from the rest of Iberia form a distinct clade that falls closest to A. (Taeniandrena) afzeliella (Kirby, 1802). These specimens come from the south of Baixo Alentejo (Cercal) and the Algarve (Aljezur, Benagil, Moncarapacho). These sequences are consistently separated from the broad A. russula clade by 5.50% (range 3.68–11.11%). Two specimens from Morocco that correspond to A. russula form a third clade of A. russula s.l. At the present time, no taxonomic action is taken; these three lineages may represent different isolated populations. More powerful genetic techniques are required to resolve this problem, as for Iberian members of the gelriae-group (see Praz et al. 2022).

Wood et al. (2021) elevated A. gredana to species status, and gave a distribution across the Sistema Central, central and northern Portugal, and northern Spain across to the Pyrenees in the province of Huesca. Examination of material from the Hautes-Pyrénées in France approximately 80 km north-east of the Huesca site revealed the presence of A. gredana in France. The species is therefore not endemic to the Iberian Peninsula, though it is likely to have a French range restricted to high altitude sites in the Pyrenees.

Material examined. France: Hautes-Pyrénées, Eget Cité, 4.v.2017, 1♂, leg. R. Rudelle, R. Rudelle Colln.; Spain: Huesca, San Juan de la Peña, 14.v.1995, 1♂, leg. H. & J.E. Wiering, RMNH (see also records in Wood et al. 2021).

This nominal species is highly variable across its range which was previously considered to be from Morocco and Iberia to Turkey and the Levant. Andrena doursana was originally described from Algeria, and Warncke (1975a, 1975b, 1980) described the subspecies A. d. citreola Warncke, 1975 from Spain, A. d. agadira Warncke, 1980 from southern Morocco, A. d. bengasia Warncke, 1980 from Libya, and A. d. mizorhina Warncke, 1975 from Turkey. Pisanty et al. (2022b) elevated the eastern subspecies A. (Truncandrena) doursana mizorhina to species status, leaving populations in Iberia and north-western Africa to Libya.

The differentiation between these remaining subspecies relies on the colour of the female pubescence, as there are no structural differences in the males; indeed, the subspecies A. d. agadira and A. d. bengasia were described only from the female sex. Andrena d. citreola is bright, and has predominantly white hairs on the face and a light brown terminal fringe with scattered white hairs laterally. Andrena d. agadira is much darker, with dark facial hairs and a uniformly dark terminal fringe. Wood et al. (2020b) found the undescribed female of a similar species endemic to Morocco, A. (Truncandrena) alchata Warncke, 1974, which is structurally very similar to the female of A. d. agadira, but these authors were unable to conclude if the two taxa were synonymous.

Analysis of barcodes from southern and northern Morocco and Iberia shows that female specimens identified as A. doursana s. str., A. d. agadira, and A. d. citreola did not form differentiated clades (Fig. 29). Light forms from southern Portugal [IBIHM1223-22] and the Moroccan Anti-Atlas near Tazenakht [WPATW505-22] were separated, and dark forms from south-western Morocco near Guelmim [WPATW506-22] and Tiznit [WPATW503-22] were clustered closer to the light specimen from Tazenakht. The light specimen from Portugal was clustered close to a specimen of A. doursana s. str. from the Middle Atlas near Taza [WPATW509-22]. This pattern strongly suggests simple separation by distance, with geographically closer specimens displaying more similar barcode sequences. Accepting a broad A. doursana species concept, these sequences showed an average intraspecific distance of 2.79% (range 0.29–5.93%).

More broadly, this A. doursana clade had bootstrap support of 78, and was sister to the A. alchata sequences generated by Wood et al. (2020b), a species that has a clearly different male morphology. Andrena alchata has bootstrap support of 91, and was separated from A. doursana by an average of 6.29% (range 5.62–8.03%). Finally, the single sequence of A. mizorhina was strongly separated from A. doursana by an average of 10.45% (range 10.18–10.94%). These genetic results justify the decision of Pisanty et al. (2022b) to elevate A. mizorhina to species level, the findings of Wood et al. (2020b) who identifed the female of A. alchata, and the original description of A. d. agadira by Warncke as a subspecies, correctly associating dark specimens from south-western Morocco with A. doursana. Given the variation in colour forms observed here, no subspecies framework is employed, and so the Iberian taxon is referred to simply as A. doursana.

Andrena medeninensis was described from Tunisia, and like A. doursana, it nominally displays great variation across its range from Morocco and Iberia to Turkey and the Levant. Warncke (1967, 1974, 1980) described several subspecies, A. m. donata Warncke, 1967 from Spain, A. m. abunda Warncke, 1974 from Morocco, A. m. tiznita Warncke, 1980 from south-western Morocco, and A. m. usura Warncke, 1967 from Turkey. Wood (2023b) synonymised A. m. usura with A. pareklisiae Mavromoustakis, 1957, leaving the North African and Iberian populations.

Sequences of A. medeninensis s. str. and A. m. tiznita formed a clade with bootstrap support of 91 (Fig. 29). However, the single available sequence of A. medeninensis s. str. differed from A. m. tiznita sequences by an average of 4.80% (range 4.70–4.85%), A. m. tiznita forming a subclade with bootstrap support of 94. However, A. m. abunda was strongly separated from this clade of A. medeninensis s.l. by an average genetic­ distance of 10.19% (range 9.97–10.57%). This A. m. abunda clade had bootstrap support of 100, and was sister to an A. medeninensis s.l. + A. tscheki Morawitz, 1872 s.l. clade. Andrena abunda stat. nov. is therefore raised to species status; it restricted to Algeria and Morocco. Morphologically, it is extremely similar to A. medeninensis s.l., but has much darker pubescence. In the female sex (the male is unknown), the only clear structural difference is that A3 exceeds the length of A4+5+6, whereas in A. medeninensis s.l. A3 is shorter than A4+5+6.

As it was unfortunately not possible to sample the Iberian subspecies A. m. donata, and no genetic sequences are available from Tunisia, the locus typicus for A. medeninensis s. str., no further taxonomic action is taken here. Given the large genetic difference displayed by A. abunda despite almost no morphological differentiation (at least in the female sex), it is difficult to comment on the Iberian subspecies which simply appears to be a colour variant of this nominally widespread species.

Andrena truncatilabris is a widespread species that was originally described from the Caucasus from what is today Armenia (Astafurova et al. 2021). The species is nominally distributed across the Mediterranean basin, from Morocco and Iberia to the Urals and Iran (Gusenleitner and Schwarz 2002). However, in the east it descends only to the Levant and does not enter the eastern part of North Africa. In the west, Warncke (1967) described two subspecies: A. t. espanola from Spain and A. t. nigropilosa from Algeria that differed from A. truncatilabris s. str. in the structure of their clypeus and their male genital capsule, though Warncke noted that the difference between the two subspecies was minimal, pointing to the darker pubescence of North African specimens as a point of difference.

Sequences from specimens from Spain and Morocco showed almost no genetic differentiation, with an average intraspecific distance of 0.99% (range 0.00–1.85%; Fig. 29). They were strongly separated from A. truncatilabris s. str. sequences from northern Israel by an average genetic distance of 9.66% (range 9.12–10.12%). Andrena truncatilabris s. str. formed a clade with bootstrap support of 99, and was sister to A. (Truncandrena) ferrugineicrus Dours, 1872, whereas A. t. nigropilosa + A. t. espanola formed a clade with bootstrap support of 100 that was sister to the A. truncatilabris s. str. + A. ferrugineicrus clade.

Given this genetic difference, it is clear that specimens from Iberia and north-western Africa are both conspecific and distinct from A. truncatilabris s. str. Given this distribution, the use of the name A. t. espanola is undesirable, and so A. nigropilosa stat. nov. is elevated to species status and A. t. espanola syn. nov. is synonymised with it as a subjective junior synonym, as the two names were described in the same publication. The updated synonymy is therefore as follows:

Andrena (Truncandrena) truncatilabris nigropilosa Warncke, 1967: 225, ♀♂ [Algeria: OÖLM, examined].

Andrena (Truncandrena) truncatilabris espanola Warncke, 1967: 224, ♀♂ [Spain: OÖLM, examined] syn. nov.

Distribution. Portugal, Spain, France, Morocco, Algeria, Tunisia (newly recorded). Material from south-eastern France and northern Italy must be carefully revised, but the position is taken here that the Maritime Alps represent a barrier between A. nigropilosa and A. truncatilabris s. str. This should be confirmed with genetic evidence.

Material examined. Algeria: S. Algeria, Laghouat [33.8082°N, 2.8316°E], iii.–iv.1929, 1♀, leg. Meyer, OÖLM (holotype of A. t. nigropilosa); Tlemcen, 20.iv.1910, 1♂, leg. de Bergeoin, OÖLM (paratype of A. t. nigropilosa); Spain: Sierra Nevada [37.0732°N, -3.3948°E], vi.1891, 1♀, leg. Handl., OÖLM (holotype of A. t. espanola); Montarco, 28.iv.1924, 1♂, leg. J.M. Dusmet y Alonso, OÖLM (paratype of A. t. espanola); Tunisia: Kef, 5 km SW Touiref, 28.iv.2012, 41, leg. C. Sevidy & A. Müller, AMC/TJWC.

Pérez (1895) described A. villipes from north-eastern Spain (Fig. 30), later describing the synonymous A. (Truncandrena) squalida Pérez, 1903 from south-western France. It has a restricted distribution, known from Cistaceae-rich habitats from northern Morocco, Portugal, Spain, southern France, and north-western Italy (Gusenleitner and Schwarz 2002; Lhomme et al. 2020). Searches in the Sierra de las Nieves in southern Spain produced two particularly large and dark specimens provisionally identified as A. villipes. A sequence from one of these specimens is separated from A. villipes sequences from southern Portugal by an average genetic distance of 11.85% (range 11.85–11.85%). This new species is described and diagnosed below.

Material examined. Spain: Barcelona [41.4028°N, 2.1332°E], 1♀, MNHN (lectotype of A. villipes; Fig. 30).