© Biologiezentrum Linz/Austria; download unter www.zobodat.at Linzer biol. Beitr. 46/2 1535-1567 19.7.2014 Andrena (Suandrena) portosanctana COCKERELL, 1922 and A. (Suandrena) maderensis COCKERELL, 1922 – new taxonomical and ecological data for two closely related endemic bee species of the Madeira Archipelago, Portugal A. KRATOCHWIL, A. SCHWABE & E. SCHEUCHL Abstract: Andrena (Suandrena) portosanctana COCKERELL, 1922, endemic to Porto Santo, Madeira Archipelago, Portugal, was described by Cockerell on the basis of three females collected in the year 1921. No further specimens were found until the year 2011, when two females of this species (collecting date 1995) were detected in a German private collection. In the year 2012 several females and, for the first time, some males were collected by two of the authors in Porto Santo (eight localities). A differential diagnosis of A. portosanctana compared to its probably closest relative A. maderensis COCKERELL, 1922 (endemic to Madeira Island) is given. Different morphological features could be demonstrated, e.g., for the following structures: length of body, wings, clypeus, pterostigma, basal area of propodeum, width of apical process of labrum basal area, puncturing of clypeus and mesoscutum, structure of apical process of labrum basal area and basal area of propodeum, colour of flagellomeres, pubescence colour of head, mesosoma, corbiculae and metasoma. It can be clearly shown that these morphological characteristics support the differentiation of two different species. Further data for A. portosanctana and A. maderensis regarding distribution ranges, habitats, flight times and specific flower-visiting behaviour as well as the role of former and recent human impact for the population size of A. portosanctana are presented. Both species are considered to be oligolectic in pollen foraging for Brassicaceae species. Aspects of the potential mainland ancestor as well as evolutionary and biogeographical aspects including other wild-bee species of the Madeira Archipelago are discussed. Given the limited pollen resources available in various Brassicaceae species, A. portosanctana is probably an endangered species with small population sizes; therefore a special conservation program should start to avoid extinction. K ey w o rd s: Andrenidae, island biogeography, species diversification, flower- visiting behaviour. Introduction COCKERELL (1922) described two Suandrena species for the Madeira Archipelago: Andrena maderensis COCKERELL, 1922 and A. portosanctana COCKERELL, 1922. In contrast to A. maderensis, which is morphologically well characterised and common on Madeira Island, until now the morphological analysis of A. portosanctana was based only on three females collected by Cockerell in January 1921 on Porto Santo Island. © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1536 No male of A. portosanctana was detected by Cockerell or anybody else, nor was there any evidence of the existence of further female individuals. Due to the unknown existence of syntypes and in absence of further specimens of this species the Andrena specialist Robert Wilhelm Grünwaldt, Munich, Germany (1909-2003), doubted the validity of A. portosanctana as distinct species (WARNCKE 1967). He argued that the short distance between Porto Santo Island and Madeira Island, only 45 km, should not serve as strong barrier for such large bee individuals. Grünwaldt had never seen the types of A. portosanctana. According to GUSENLEITNER & SCHWARZ (2002) A. portosanctana is a synonym for A. maderensis. This is also argued in the Fauna Europaea (DE JONG 2013). In contrast, WARNCKE (1967) gave A. portosanctana the status of a subgenus (A. m. portosanctana COCKERELL, 1922), according to the oral contribution of Grünwaldt. WARNCKE (1968) defined A. m. maderensis COCKERELL 1922, endemic to Madeira Island, as nominate species. Besides A. m. portosanctana (Porto Santo), WARNCKE (1968) considered two other taxa as subspecies of A. maderensis: A. m. notata WARNCKE, 1968 (Canary Islands) and A. m. fratella WARNCKE, 1968 (Morocco); see also GUSENLEITNER & SCHWARZ (2002). DYLEWSKA (1983) objected in analogy to Grünwaldt, that types of A. maderensis do not exist; in her opinion a differentiation in the subspecies A. m. maderensis and A. m. portosanctana is not possible without ana- lysing the types. In the course of our studies we detected two of the three syntypes of A. portosanctana. Syntypes of females have been designated, because no holotype was dedicated in the species description. The first syntype is deposited in the California Academy of Science, California, San Francisco, USA (CAS TYPE 15373: syntype, female, adult), the second one in the Department of Entomology, NMNH, Smithsonian Institution, Washington DC, USA (USNM 24656, barcode number 00533697). Here we present both syntypes of A. portosanctana by photos in lateral view, dorsal view, head profile and by labels (Fig. 1- 8). We were not able to discover where the third female is (was) deposited. Madeira Island (728 km²) is inhabited by 16 wild-bee species (eight endemic, three native, five introduced or not permanently established) in contrast to Porto Santo Island (42 km²), where eight bee species (five endemic, three native, one introduced) were detected (KRATOCHWIL et al. 2008). Some species occur on both islands, e.g., Amegilla quadrifasciata maderae (SICHEL, 1868), Bombus maderensis ERLANDSSON, 1979, Lasioglossum villosulum (KIRBY, 1802), Lasioglossum wollastoni (COCKERELL, 1922). But there are bee species with a remarkable allopatric distribution, if Madeira Island and Porto Santo are compared, although the distance of 45 km between these islands should usually not act as a strong dispersal barrier, especially for larger bee species (e.g. species with more than 10 mm body length). But the different age and colonization history of the two islands (Porto Santo 14 million years old, Madeira 4.6 million years old; GALOPIM DE CARVALHO & BRANDÃO 1991, GELDMACHER et al. 2000) and the trade winds blowing from the Northeast (preventing eastward dispersal), could be arguments for the occur- rence of two different endemic Suandrena species. KRATOCHWIL & SCHEUCHL (2013) documented recently that A. wollastoni COCKERELL, 1922 (Madeira Island) differs from A. dourada KRATOCHWIL & SCHEUCHL, 2013 (endemic to Porto Santo) in many morphological features (females: e.g. length of body, wings, clypeus, stigma, propodeum; facial fovea index; pubescence colour of, e.g., © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1537 paraocular area, tibial scopa, tergites 5 and 6, labrum process structure, propodeum sculpture; males: e.g. length of body, wings, clypeus, stigma; width of labrum process; propodeum sculpture). A. dourada is a species of its own, formerly grouped to A. wollastoni. Another example is Osmia madeirensis VAN DER ZANDEN, 1991 (endemic to Madeira Island) and O. latreillei iberoafricana (PETERS, 1975) (native in Porto Santo, but also in mainland Europe and also in North-Africa). We hypothesise the occurrence of another species pair: Andrena maderensis COCKERELL, 1922 (endemic to Madeira Island) and A. portosanctana COCKERELL, 1922 (endemic to Porto Santo). 89 years after the first description by COCKERELL (1922) two females of A. portosanctana were detected in the year 2011 in the private collection of Christoph Saure (Berlin, Germany). The two females were collected by Gerhard Jaeschke (Berlin), a radiologist and ambitious hobby entomologist (KLIMA 1995). Both females were labelled by Fritz Gusenleitner (Linz, Austria) as Andrena cyanomicans PÉREZ, 1895, and this information was published by FELLENDORF et al. (1999). Erwin Scheuchl identified the specimens as A. portosanctana in November 2011; this is in total congruence with the species description by COCKERELL (1922). In view of these findings we increased our efforts to study A. portosanctana by searching for further females and the so far unknown male. We were successful in detecting 19 females and 5 males in the year 2012 and managed to study the flower-visiting behaviour of some of these individuals. First we will characterise A. portosanctana concerning morphology of females and males, distribution, habitats, flight time and flower-visiting behaviour. According to the hypothesis of the close relationship between the two species, we will subsequently ana- lyse A. maderensis in the same way. Additionally we will give ecological characterisa- tions (habitats, flight times, flower-visiting behaviour). A differentiating diagnosis will follow for the two species. Thereafter we will discuss the relationship of both species to other Suandrena species, and furthermore add some biogeographical aspects and con- sider origin and relationship of A. portosanctana and A. maderensis in combination with the evolutionary background. The role of former and recent human impact regarding the population size of A. portosanctana and aspects of nature conservation will also be discussed.  Material and methods The Madeira Archipelago is situated above the Azores Ridge 32.30° N to 33.30° N and - 16.30° E to -17.30° E in the Atlantic Ocean, all about 560 km away from Morocco and 978 km from Lisboa (Portugal). It includes Madeira Island (728 km²), Porto Santo Island (42 km², 45 km northeast), Ilhas Desertas (three small islands, 24 km westward) and some other small islands. According to GALOPIM DE CARVALHO & BRANDÃO (1991) and GELDMACHER et al. (2000) the age of the Madeira Archipelago ranges from 3.6 to 14 million years (Porto Santo: 14, Madeira: 4.6, Desertas: 3.6). During the last glacial period (optimum 18,000 years BP) Madeira and the Desertas were probably connected by a landbridge (BREHM et al. 2003) (today depth of the sea about 90 m). During the past millions of years the dis- tance from the European continent was influenced by processes of continental drift, sea- level variations and geological processes of emergence and erosion of landbridge islands. In most of all cases sea-level variations represent the most important influencing factors. © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1538 Reconstructions of the glacial coast line during the last glacial optimum (18,000 years BC) demonstrate that the sea was 120 m deeper than today. Furthermore, many islands covered larger areas than today and were situated above sea-level (currently below sea- level); see GARCIA-TALAVERA (1999). MOORE et al. (2002) thus assume a minimal dis- tance of 200 km from the European continent in former times. Madeira Island is very mountainous and rich in barrancos; the highest elevation is the summit of ‘Pico Ruivo de Santana’ (1862 m a.s.l.). The coast lines are generally steep. In contrast the island of Porto Santo is much more flat; the highest peak ‘Pico do Facho’ reaches 517 m a.s.l. In contrast to Madeira Island Porto Santo has a band of sandy flat coastal line 10 km long. The highest point of Deserta Grande is unnamed (461 m a.s.l.); the coastal slopes are steep. The Madeira Archipelago belongs to the Biogeographic Region ‘Mediterranean’, Subre- gion ‘Madeiran’; RIVAS-MARTINEZ et al. (2004). The climate is oceanic, and mostly there are trade winds blowing from northeast. This is the reason why in the northern mountainous areas of Madeira Island the precipitation is higher than in the southern areas. On windward sides the annual precipitation amounts about 3000 mm per year; on lee sides, the precipitation is about 500 to 600 mm per year. The southern coastline and the eastern coastline areas (Ponta de S. Lourenço) are very dry. The mean temperature of Funchal in January is about 16 °C, in July (highest values) nearly 30 °C. The highest values of precipitation are obtained in Funchal from November to February (TAVARES 1965). The oceanic Mediterranean-Madeiran climate allows the existence of laurel forests (Laurisilva). Some of the plant species of Laurisilva are thought to be relicts from the flora of the tertiary period (POTT 2005). In the case of Porto Santo, the altitudes of the mountains mainly do not reach the trade wind zone, therefore for main parts of the island the humid clouds pass away. In times of the discovery of Madeira Archipelago, Madeira was almost completely covered by forests or shrubs, which is documented by the florentine sea map of the year 1351 (Medici-Laurentian Atlas), where the Madeira island was named ‘Isola di Legname’ = ‘woody island’. In the report of the discoverer João Gonçalves Zarco (* about 1380; † about 1467) arboreous vegetation was mentioned for the coastal line to the mountain peaks (year 1419). Nowadays the natural vegetation is partly destroyed, frag- mented or replaced by secondary vegetation. The climate of Porto Santo is much drier compared to Madeira Island. The potential natural vegetation with Dracaena draco subsp. draco (extinct), Juniperus turbinata subsp. canariensis (rare), Erica platycodon subsp. maderinicola (rare), Olea maderensis (rare), probably with spotty former distribution of Sideroxylon mirmulans (rare) and Apollonias barbujana (extinct) had been destroyed over wide areas (rare/extinct data according to PRESS & SHORT 1994), in addition, the former soils eroded in a broad scale (FAUST-LICHTENBERGER 1988). In the 15th and 16th centuries Porto Santo was intensively cultivated after clearing (agriculture, e.g. crops, and livestock breeding) because the island served as one of the most important areas for supplies of food for ships. With the end of the 16th century many inhabitants left the island as consequence of the frequent attacks of buccaneers. The agricultural landscape was devasted and erosion and degrada- tion processes started. This was the beginning of an immense loss of synanthropic vege- tation types. In the 18th century only a few farmers lived there, in great misery. After reforms of the Portuguese government, coordinated by the Marquis of Pombal, Sebastião © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1539 José de Carvalho e Mello (1699-1782), the conditions became more favourable to settle as a farmer on Porto Santo Island. Today, ruins of old windmills, terraces and boundary stone walls for erosion shelter are documents of this time. But erosion is an insurmount- able obstacle. Since the sixties of the last century agriculture was extremely reduced. Soil erosion control and recovery of endogenous biodiversity are amoung the main points for municipality, agricultural and environmental entities (DI BERARDINO 2010). The specimens studied were collected by Anselm Kratochwil, Angelika Schwabe and Claudius Kratochwil (hand netting, pantraps white, yellow, blue) in the years 1995 (31.03.-13.04.), 2005 (24.03.-11.04.) and 2012 (15.03.-22.03.). Concerning A. portosanctana we were able to analyse 19 females (18 newly trapped in the year 2012 and one female from the collection of C. Saure) and 5 males (newly trapped in the year 2012), in the case of A. maderensis we collected 51 females and 10 males in the years 1995 and 2005 on Madeira Island; 3 further studied males come from the collection of Dr. Herbert Hohmann (‘Übersee-Museum’ Bremen, Germany). We analysed various morphological features, e.g., colour and structure of different body parts including pubescence (head: e.g. clypeus, labrum process, antennae, paraocular area, fovea facialis, genae); mesosoma (e.g. mesothorax, mesoscutum, scutellum, basal propodeum, femur, tibia, basiotarsus, mediotarsi, propodeal corbiculae, trochanteral and femoral flocculus, tibial scopa, wings, pterostigma); metasoma. The following morphometric features were analysed (terms according to MICHENER 2007, TADAUCHI & XU 1995, ARIANA et al. 2009, KRATOCHWIL & SCHEUCHL 2013): body length (BL): in mm from antennal base to tip of pygidium; wing length (WL): length of forewing excluding tegula; tergites 1-5 (T1-5); head length (HL): from top of vertex to lower margin of clypeus; head width (HW); mesosomal width (MsW): between outer rims of tegulae; metasomal width (MtW): maximum width of terga from dorsal view; ocellocular distance (OOD); postocular distance (POD); ocelloccipital distance (OCD), maximal length of facial fovea (FVL); maximal width of facial fovea (FVW); length of flagellomeres 1-3 (FL1-3): measured on ventral surfaces of flagellomeres when antenna stretched forward; puncture diameter (Pd); distance to a nearby puncture (IS); clypeal length (CPL); labrum process (apical process) width at the top (LBW); pterostigma length (PSl); propodeum basal area length (PBAl); SD = standard deviation. According to MICHENER (2007), the terminology for parts of the labrum is often confused in the genus Andrena (especially the term ‘process’ for the ‘basal elevated plate’). He wrote: ‘The term process is misleading because this plate does not project, as one expects of a process. In other bees, e.g. the Panurginae (see RUZ 1986), the same structure is called the basal area of the labrum. Use of the word ‘process’ in the sense of basal area is further confusing because in some bees, especially the Halictidae, there is an entirely different process on the apex of the labrum, here called the apical process of the labrum.’ In the following we differentiate the labrum in a) basal area, b) apical (distal) process of the basal area, c) vertical area, d) margin of the basal area (Fig. 9). If an apical process is absent, the basal area can be developed broadly as a semicircle (e.g. female of A. rufizona IMHOFF, 1834) or trapezoid (female of A. bimaculata [KIRBY, 1802]). In many cases the basal area is more or less narrow (female of A. minutuloides PERKINS, 1914; female of A. barbilabris KIRBY, 1802). The fact that there are many examples of such a narrower basal area had led to the synonymous use of the term ‘pro- © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1540 cess’. In some cases an apical process is well developed and the basal area can be differ- entiated by a characteristic form of margins (e.g. female of A. coitana [KIRBY 1802]). If the margin has an irregular shape it is difficult to detect the apical process of the basal area (e.g. female of A. lathyri ALFKEN, 1899). The form of the apical process can be also characterised as e.g. square-based, trapezoid or ligular, as emarginated or not. In A. maderensis the apical process is well developed. We measured the apical process width at the top. In A. portosanctana an apical process is missing or extremely frag- mented. We measured the length of the apical part of the trapezoid labrum, which is as a rule the base of the apical process. A special type is found e.g. in females of A. labiata (KIRBY, 1802) or females of A. ovatula (KIRBY, 1802) where the distal end of the basal area is emarginated but not exposed. This is effected by three separated hollows with hairs. Two ‘bridges’ of the basal area situated between the hollows reach into the vertical area and are more or less hairless. A similar situation is given in the case of the males of A. maderensis and pars parte of A. portosanctana. Here we measured the width of the emarginated area including the width of the both ‘bridges’. Boxplots were constructed by Excel 2010. The bottom and top of the box are the 25th and 75th percentile (lower quartile and upper quartile), and the band near the box is the median (50th percentile), the point is the arithmetic average. The ends of the vertical lines or ‘whiskers’ indicate the minimal and maximal data values. The nomenclature for plant species names follows JARDIM & SEQUEIRA (2008). One plant species was newly described in the year 2010: Echium portosanctensis CARVALO, PONTES, BATISTA-MARQUES & JARDIM. Species descriptions 1. Andrena (Suandrena) portosanctana COCKERELL, 1922 A. maderensis portosanctana COCKERELL, 1922 S t at u s: The first who discussed the question of the status of A. portosanctana after COCKERELL (1922) was WARNCKE (1967). He referred to an oral communication of Grünwaldt and noted that A. portosanctana is only a subspecies of A. maderensis. Neither Warncke nor Grünwaldt, ever saw specimens of A. portosanctana. Former Descriptions In the first description by COCKERELL (1922), females are only differentiated on the basis of some morphological features: ‘Female. – Like Andrena maderensis, and differing similarly A. bimaculata, but the thin abdominal hair-bands are white, without any fulvous tint, the apical tuft is black; the wings are clearer, not so red; the hair of front and vertex (but not occiput) is black, and on face dull white, but there is a conspicuous reddish band from eye to eye at level of antennae; the discs of mesothorax and scutellum have pure black hair, that on pleura is long and white; the process of labrum is more rounded, without a distinct point or tubercle; the clypeus, though shining, lacks a distinct smooth median line. The greenish tint of the abdomen is very obscure.’ © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1541 POPOV (1958) published an identification key for species of the subgenus Plastandrena and incorporated erroneously A. maderensis and A. portosanctana at the basis of Cockerell’s description in this subgenus. Up-to-date description Female. BL 12.27 mm (SD 0.42), WL 8.77 mm (SD 0.42). C o l o u r . Head (Fig. 3, 7): black, part slightly greenish; flagellomeres 3-10: upper side brownish, lower side yellow; mandible black, as a general rule with a reddish tip. Mesosoma (Fig. 1, 2, 5, 6): black; femur and tibia 1 and 2 black, tibia 3 sometimes pars parte reddish brown; metatarsi 1 and 2 brown, metatarsus 3 pars parte reddish-brown, mediotarsi black, pars parte reddish brown; wings hyalin, veins brown (reddish/black), pterostigma orange (yellowish) with a dark margin. Metasoma (Fig. 2, 6): T1-4 black partly greenish (see description by COCKERELL 1922), with black to dark reddish brown depression zones; pygidium black. Pubescence. Head (Fig. 3, 7): white hairs in front, yellowish hairs beside and between the antennal sockets (see description by COCKERELL 1922: ‘there is a conspi- cuous reddish band from eye to eye at level of antennae’; Fig. 7); upper part of paraocu- lar area with brownish hairs; clypeus as a general rule only with laterally denser yellowish-white hairs; scapus with yellowish hairs and some brownish hairs; genal area with dense whitish hairs (in many cases brighter than the paraocular hairs), Fig. 5; lower part of facial fovea with whitish hairs, upper part with brownish hairs; vertex with some dark brownish (black) hairs. Mesosoma: mesoscutum and scutellum with lateral yellowish-white hairs (Fig. 2, 6), in the centre only some brownish hairs (only in some cases black hairs; compare the description by COCKERELL 1922); mesepisternum with longer yellowish-white hairs (Fig. 1, 5); propodeal corbicula with dense white marginal hairs, no hairs in the centre; trochanteral and femoral flocculus well developed with long white hairs (Fig. 1, 5); dorsal tibial scopa with reddish-brown hairs and brownish tips, dorsobasal tibial scopa with brownish hairs (Fig. 5); ventral tibial scopa with yellowish whitish hairs (Fig. 5). Metasoma (Fig. 1, 2, 5, 6): T1-4 with white partly closed thin hair bands, T5 in the centre with dark brownish hairs, lateral with long white hairs, T6 with dense dark brownish hairs reaching to pygidium, centre of T1 and T2 with longer, T3-4 with shorter whitish hairs. S t r u c t u r e . Head (Fig. 3, 7): HL/HW = 0.83 (SD 0.02); HW:MsW:MtW = 1.1:1.1:1.0; vertex narrow, as wide as ocellar diameter, surface structure not rugose; face above antennal fossae with longitudinal rugulae, interrugal space slightly shiny; OOD:POD:OCD = 3.5:2.7:1.0; FL1:FL2:FL3 = 2.4:1.1:1.0; inner margin of eyes not converging; FVL = 1.36 mm (SD 0.08), FVW = 0.31 mm (SD 0.03), FVL/FVW = 4.39 (SD 0.30); facial fovea at the base narrower; clypeus convex, front very shiny without punctures, other area also shiny and smooth, as a general rule without impunctate median line (see the description of COCKERELL 1922), punctured, from the centre to the base shagreened; shallow punctures (Pd = [14] 28-42 μm, IS = 14-56 [70] μm); CPL = 1.26 mm (SD 0.07); labrum with basal area margin rounded, as a general rule slightly trapezoid (sometimes undulated or curled) contrary to the description of COCKERELL 1922: ‘the process of labrum is more rounded, without a distinct point or tubercle’) mostly with a fragmented apical process (sometimes asymmetric and/or emarginated), vertical area with long yellowish/reddish hairs; LBW = 1.26 mm (SD 0.02). Mesosoma © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1542 (Fig. 2, 6): shagreened, shallow punctured (Pd = 28 μm); PSl = 1.25 mm (SD 0.07); rugose, as a general rule some longitudinal lamina, small central lateral boundary line, PBAl = 0.50 mm (SD 0.03). Metasoma (Fig. 2, 6): smooth and shiny, scattered punctured (Pd = 14-28 μm), posterior depression of T1-T5. Male. BL 9.99 mm (SD 0.58), WL 7.77 mm (SD 0.33). C o l o u r . Head: black, part slightly greenish; flagellomeres 3-10: lower side brownish, upper side black/dark brownish; mandible black with a reddish tip. Mesosoma: black; femur and tibia 1 and 2 black, tibia 3 pars parte reddish brown; metatarsi 1 and 2 brown, metatarsus 3 pars parte reddish-brown, mediotarsi black, pars parte reddish brown; wings hyalin, veins brown (reddish/black), pterostigma orange (yellowish) with a dark margin. Metasoma: T1-4 black partly greenish, with black to dark reddish brown depression zones; pygidium black/reddish. Pubescence. Head (Fig. 12): white hairs in front, yellowish hairs beside and between the antennal sockets, upper part of paraocular area with brownish hairs; clypeus with whitish lateral hairs, hair free line in the centre; scapus with whitish hairs, brownish hairs behind antennal socket; dense whitish hairs, brighter than paraocular hairs, upper area with brownish hairs; vertex with dark brownish (black) hairs. Mesosoma (Fig. 12): mesoscutum and scutellum with whitish hairs lateral, in the centre only few hairs, some brownish; mesepisternum with long whitish hairs. Metasoma (Fig. 12): T2-4 with whitish thin hair bands, in the centre slightly open, T6 central with reddish, lateral whitish hairs, T7 with reddish hairs, in the centre T1 and T2 with long, T3-4 shorter whitish hairs; T8 narrow with white hairs. S t r u c t u r e . Head: HL/HW = 0.81 (SD 0.04); HW:MsW:MtW = 1.2:1.1:1.0; vertex and face above antennal fossae similar to female; OOD:POD:OCD = 3.5:2.6:1.0; FL1:FL2:FL3 = 1.5:1.1:1.0; inner margin of eyes not converging; vertex narrow, as wide as ocellar diameter; clypeus structure similar to female (Pd = 28-42 μm, IS = 14-42 μm), CPL = 1.05 mm (SD 0.04); labrum basal area flat and rounded, without exposed apical process, beneath in the centre a hollow with whitish/yellowish hairs, left and right with a hairless area, followed by an area with long yellowish hairs; width of the emarginated area including the width of the two ‘bridges’ = 0.21 mm. Mesosoma: mesoscutum and scutellum shagreened, scattered shallow punctured (Pd = 28 μm); PSl = 1.07 mm; propo- deum rugose in the centre, no lateral boundary line; PBAl = 0.42 mm (SD 0.04). Metasoma: smooth and shiny, very scattered punctures (Pd = 14-28 μm), slightly shagreened, posterior depression of T1-T5. Genitals: Fig. 13, 14. S pecimens examined (identity number, sex, locality, altitude above sea level, latitude and longitude coordinates, flower-visiting behaviour, date of collection; PS: private collection A. Kratochwil, CS: private collection C. Saure): Females: PS12/29-PS12/34: 6(cid:2)(cid:2), Porto Santo, east of Campo de Baixo, south of restaurant ‘Mare Sol’, dune area, 5 m, 33°02'53.33''N, - 16°20'50.46''E, Cakile maritima subsp. maritima, 17.03.2012; PS12/53, PS12/60, PS12/65, PS12/69, PS12/70, PS12/72, PS12/73: 7(cid:2)(cid:2), Porto Santo, Campo de Baixo, Estrada dos Carreiros, dune area, 5 m, 33° 02'45.58''N, -16°21'0.22''E, Cakile maritima subsp. maritima, 18.03.2012; PS12/82: 1(cid:2), Porto Santo, Vereda do Pico Branco, fallow land, 310 m, 33°05'31.84''N, - 16°18'17.16''E, Calendula arvensis, 19.03.2012; PS12/142: 1(cid:2), Porto Santo, Vila Baleira, near sports field, ruderal site, 25 m, 33°02'35.16''N, -16°21'38.48''E, Asphodelus fistolosus, 20.03.2012; PS12/150, PS12/151, PS12/153: 3(cid:2)(cid:2), Porto Santo, Campo de Baixo, near tennis court, ruderal, fallow land, 20 m, 33°02'50.07''N, -16°21'23.35''E, Sinapis arvensis, 20.03.2012; PS12/154: 1(cid:2), Porto Santo, Capela de S. Pedro, ruderal site, 50 m, 33°02'44.85''N, -16°21'43.82'', Rapistrum © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1543 rugosum, 20.03.2012; CS1: 1(cid:2), surrounding Hotel Luamar, dune area, 5 m, 33°02'12.80''N, 16°21'36.86''E, 18.03.1994, leg. G. Jaeschke. Males: PS12/35: 1(cid:3), Porto Santo, east of Campo de Baixo, south of restaurant ‘Mare Sol’, dune area, 5 m, 33°02'53.33''N, -16°20'50.46''E, Cakile maritima subsp. maritima, 17.03.2012; PS12/41M: 1(cid:3), near Pico da Cabrita, fallow land, 230 m, 33°05'42.18''N, -16°19'04.49''E, 18.03.2012; PS12/66-PS12/68, PS12/74: 4(cid:3)(cid:3), Campo de Baixo, Estrada dos Carreiros, dune area, 5 m, 33°02'45.58''N, -16°21'0.22''E, Cakile maritima subsp. maritima, 18.03.2012. Historical records  COCKERELL (1922): 3(cid:2)(cid:2) collected in January 1921, Porto Santo, in the south of Pico do Castelo on flowers of Oxalis pes-caprae and Calendula.  FELLENDORF et al. (1999): 2(cid:2)(cid:2) (collection of C. Saure, Berlin) were collected by G. Jaeschke (Berlin) in the surrounding of Hotel Luamar (18.03.1994) determined by F. Gusenleitner (Linz) as Andrena cyanomicans. Both were identified as A. portosanctana (see Introduction).  FELLENDORF et al. (1999): 3(cid:2)(cid:2), visiting flowers of the Brassicaceae taxa Cakile maritima and Raphanus sp. (18.03.-30.03.1997), were identified by the authors as A. cyanomicans. According to FELLENDORF et al. (1999) 1(cid:2) had been deposited in the Bavarian State Collection of Zoology (Munich). Unfortunately the specimen sent to the Bavarian State Collection was lost before deposition by the courier (Schönitzer, Bavarian State Collection of Zoology Munich, pers. communication). It is quite evident that FELLENDORF et al. (1999) had not taken the original description of COCKERELL (1922) as a basis for the identification of A. portosanctana. They wrote: ‘We were unable to collect the species [A. portosanctana] on the island, despite careful search.’ Distribution and habitat characteristics 25 individuals of A. portosanctana were detected on more than one-third of all investi- gated localities (eight from 21 localities, distributed all over the island). Therefore we can assume that A. portosanctana is widely distributed on Porto Santo. The range of altitudes reaches from sea level (drift line) up to 310 m a.s.l. Individuals of A. portosanctana were found primarily in the infra-Mediterranean-Madeiran and lower thermo- Mediterranean-Madeiran thermotypes (terminology according to CAPELO et al. 2004). Most individuals (n = 18) were detected in the dune zone with occurrence of Cakile maritima subsp. maritima (Fig. 15, 16), but also on dry rocky grassland at the base of Pico da Cabrita (230 m a.s.l.) (Fig. 17), on formerly cultivated sites (Pico Branco, Fig. 18); ruderal sites or cultivated field margins near Vila Baleira and Campo de Baixo (Fig. 19, 20). Flight time records As yet flight activity has been detected from January (observation of Cockerell) to the end of March. Flower-visiting behaviour Flower visits by females and males have been observed by A. Kratochwil and A. © Biologiezentrum Linz/Austria; download unter www.zobodat.at 1544 Schwabe on individuals of five plant taxa belonging to four plant families: Brassicaceae: Cakile maritima subsp. maritima (Fig. 11): 13(cid:2)(cid:2), 6(cid:3)(cid:3) patrouilling on flowers; Sinapis arvensis: 3(cid:2)(cid:2); Rapistrum rugosum: 1(cid:2), Asphodelaceae (Liliaceae s.l.): Asphodelus fistulosus: 1(cid:2); Asteraceae: Calendula arvensis: 1(cid:2). Additionally there is the observation of 3(cid:2)(cid:2) made by COCKERELL (1922): Oxalidaceae: Oxalis pes-caprae: 1(cid:2); Asteraceae: Calendula arvensis: 1(cid:2); 1(cid:2) not assigned. The Brassicaceae-dominated flower visits of pollen-collecting females is consistent with the hypothesis, that the species of the subgenus Suandrena prefer this family (DYLEWSKA 1983, KRATOCHWIL 1991). Flower-visiting behaviour in an evolutionary, historical and biogeographical context The currently known pollen resource spectrum of A. portosanctana is dominated by flower visits on Cakile maritima subsp. maritima. This annual species grows in sandy driftline habitats. The fruits (distal segments) can float even for longer distances and periods in seawater. The seeds from the proximal segments are dispersed by wind (DAVY et al. 2006). Within the Madeira Archipelago sandy coastal habitats exist only in Porto Santo; Cakile maritima subsp. maritima is restricted to Porto Santo (PRESS & SHORT 1994). These sediments are of quarternary origin, where the sand was blown out from the large shelf of the island, which had become dry during glacial times. Eolianites (dunes of crossbedded marine bioclastic calcarenite, age: 21.570 and 13.480 B.P., main Würm), reach up to a thickness of 50 m. Eolianites are rich in fossil land snails (LIETZ & SCHWARZBACH 1971). The colonization by the ancestor of A. portosanctana probably took place much earlier than the development of the dunes and the sandy coast (including the habitat of Cakile maritima subsp. maritima). Therefore the question arises of poten- tial pollen resources within the Brassicaceae especially in former times (probably since Miocene). There are three endemic plant species of Brassicaceae which occurred in Porto Santo Island before human settlement and still exist now, but today mostly in very reduced individual numbers. - Erysimum arbuscula. The habitats of this endemic species are rock crevices on the peaks and higher slopes from 300-450 m a.s.l. E. arbuscula is a member of the Siderito multiflorae-Echietum portosanctensis, a nanophanerophytic community characterised by further endemic species (e.g. Echium portosanctensis). The commu- nity occurs in the vertical cliffs of the Pico Branco in the northern slopes of the island (Mediterranean xeric-oceanic, infra-Mediterranean dry climate) (CARVALHO et al. 2010, COSTA et al 2012). According to PRESS & SHORT (1994) E. arbuscula was for- merly locally abundant but now reduced by overgrazing in places accessible for sheep and goats. This species is also an element of the Erysimo arbusculae-Artemisietum argenteae (CAPELO et al. 1999), but only in primary habitats (CAPELO et al. 2003). - Matthiola maderensis. The habitats are costal rocks and cliffs, ‘fairly frequent in Porto Santo and sometimes found inland, recorded growing at altitudes of up to 200 m on Pico do Castelo’ (PRESS & SHORT 1994). M. maderensis is an element of