Vol. 55 No. 2 2011 Journal of Apicultural Science 131

Homology of Submarginal Crossveins in Forewings of Bees (: Apiformes)

A d a m Tofilski

Department of Pomology and Apiculture, Agricultural University, 29 Listopada 54, 31-425 Krakow, Poland, tel: +48 12 6625069 e-mail: [email protected] Received 10 August 2011; Accepted 25 November 2011

S u m m a r y In the forewings of bees (Apiformes) there are either two or three submarginal cells. In the case of wings with two submarginal cells it is not clear if the remaining crossvein separating the cells is homologues with the first or the second submarginal crossvein. Information about homology of the crossveins is important for reconstructing phylogenies of bees. I attempted to determine the homology of the submarginal crossveins using quantitative methods. Coordinates of 14 vein junctions were superimposed to determine the expected position of the submarginal crossveins. It was expected that distribution of the position of the remaining crossveins in wings with two submarginal cells would be similar to the combined distribution of the first and second crossvein in wings with three submarginal cells. However, the distributions differed markedly between wings which had two or three submarginal cells. The position of the remaining crossvein was often between the expected position of the first and second submarginal crossvein. Moreover, the distribution of the remaining crossvein was bimodal, which confirms earlier suggestions that both the first and second submarginal crossvein may be lost in evolution. The results suggest that in bees there are patterns of wing venation which are preferred by natural selection. Similar patterns can be a result of the disappearance of either the first or second submarginal crossvein. This makes the reconstruction of remaining crossvein homology difficult. Keywords: bees, Apiformes, wing, venation, submarginal crossvein, homology.

Introduction crossvein”. This terminology should Almost all bees (Apiformes) share a not be misleading, as I do not take into characteristic venation pattern. Only in consideration here any other crossveins Meliponini, Neolarrini, Allodapini, and than the submarginal. In older literature, the Euryglossinae can the venation be reduced crossveins were called transverse cubital at the distal part of the wing (Michener, veins (Robertson, 1926) or radiomedial 2000). Within the characteristic pattern, veins (Louis, 1973). there are two major types of venation with In evolution, the submarginal crossveins two or three submarginal cells. Those two were lost independently in many patterns result from the loss of either the systematic groups. In at least 6 families second abscisa of Rs (2nd Rs) or first r-m (Andrenidae, , Colletidae, (1r-m). Following terminology of , Megachilidae, Melittidae) Michener (2000), the two veins are there are species with both two and three called here; first and second submarginal submarginal cells. Even in some genera crossveins. For the sake of brevity in this (e.g. Epeolus, Eucera, Lasioglossum, paper, I often skip the word “submarginal” Leioproctus, Nomada, Rhopalolemma, and use the terms: “first crossvein” and Sphecodosoma) there are species with two “second crossvein”. In the case of wings or three submarginal cells. Within species, with two submarginal cells, the crossvein all individuals usually have either two or separating the cells is called the “remaining three submarginal cells. Only occasionally 132 there are individuals with unusual venation. the homology of submarginal crossveins In bees with three submarginal cells (e.g. in forewings of bees. Using the position of Andrena, Halictus, Lasioglossum, Melecta, some veins (homology of which is known), Nomada, Sphecodes), either the first or the position of the remaining crossvein was the second crossvein was found missing calculated and compared with the expected (Peters, 1969). In honeybees, the second position of the first and second crossvein. submarginal crossvein is sometimes It was expected that the distribution of the incomplete (Alpatov, 1928). In Eucera position of the remaining crossveins would and Pseudopanurgus, there are normally be similar to the combined distribution of two submarginal cells but occasionally the first and second crossveins. there are three of them (Robertson, 1922; Peters, 1969). The unusual venation is Material and Methods more common in males (Peters, 1969). In the analysis, forewings of 119 species Michener (2000, p. 105) suggested that of bees were used. The species represent a a submarginal crossvein can disappear in wide range of systematic groups: 7 families, a taxonomic group because some genes 19 subfamilies and 115 genera. No more are inactivated and later in evolution the than one species was used from one crossvein can appear again when the genes subgenus. There were 47 wings with two are activated. It is not clear if the number submarginal cells (Tab. 1), and 72 wings of submarginal cells has any adaptive with three submarginal cells. The analysed value. It has been suggested, that the species with three submarginal cells loss of the crossveins can be related to were: Agapostemon texanus Cresson, smaller size (Peters, 1969) and parasitism Ancyloscelis panamensis Michener, (Robertson, 1926). Andrena (Callandrena) accepta When there are two submarginal Viereck, Anthophora occidentalis cells, it is often not clear which of the Cresson, Anthrenoides meridionalis two submarginal crossveins is missing. (Schrottky), Apis mellifera Linnaeus, This information could be important for Augochlora pura (Say), Augochlorella the reconstruction of phylogenies. In the striata (Provancher), Augochloropsis past, the relative size of submarginal cells metallica (Fabricius), Bombus was used to predict which crossvein is pennsylvanicus (Degeer), Cadeguala missing. If the first submarginal cell was albopilosa (Spinola), Caenonomada significantly longer than the second, it was bruneri Ashmead, Callomelitta picta suggested that the first crossvein was lost. Smith, Canephorula apiformis (Friese), However, in some cases the difference Caupolicana hirsuta (Spinola), Centris between the submarginal cells is small. (Centris) poecila Lepeletier, Ceratina Micherner (2000) suggested that in dupla Say, Chalepogenus caeruleus Hylaeinae, in most cases, the first crossvein (Friese), Chrysocolletes moretonianus is missing. In some genera, for example (Cockerell), Colletes sp. Latreille, Hyleoides, it is impossible, however, to Deltoptila montezumia (Smith), determine which crossvein is missing. On Diadasia afflicta (Cresson), Dieunomia the other hand, it was suggested that in nevadensis (Cresson), Diphaglossa most Panurginae, the second submarginal gayi Spinola, Doeringiella (Triepeolus) crossvein was lost (Robertson, 1922). verbesinae (Cockerell), Epeoloides The methods used so far for determination coecutiens (Fabricius), Epeolus cruciger of homology of the remaining crossvein Panzer, Epicharis (Epicharana) elegans were to a large degree, arbitrary, and were Smith, Euglossa cordata (Linnaeus), applicable only in rare occasions when the Exaerete smaragdina (Guérin-Méneville), difference between submarginal cells was Exomalopsis (Exomalopsis) zexmeniae large. The aim of this study was to use Cockerell, Habralictus trinax (Vachal), quantitative methods in order to determine Homalictus dampieri (Cockerell), Vol. 55 No. 2 2011 Journal of Apicultural Science 133

T a b l e 1 . Classification of the remaining crossvein in forewings of bees as the first or second submarginal crossvein; classification should be interpreted carefully especially in cases where posterior classification probability P is low. Classification Family Subfamily Species of crossvein P Leioproctus (Filiglossa) filamentosus (Rayment) first 0.993 Coletinae Scrapter nitidus (Friese) second 0.509 Xeromelissinae Chilicola (Anoediscelis) ashmeadi (Crawford) second 0.999 Hylaeus (Heterapoides) extensus (Cockerell) second 0.977 Hylaeinae Colletidae Hyleoides concinna (Fabricius) first 0.972 Euhesma goodeniae (Cockerell) second 0.997 Euryglossa subsericea Cockerell second 0.605 Euryglossinae Hyphesma atromicans (Cockerell) second 0.999 Xanthesma furcifera (Cockerell) second 0.999 Calliopsis (Calliopsis) andreniformis Smith first 0.982 Calliopsis (Hypomacrotera) subalpina Cockerell first 0.753 Callonychium minutum (Friese) first 0.795 Camptopoeum friesei Moc sár y first 0.999 Macrotera (Macrotera) bicolor Smith second 0.997 Andrenidae Panurginae Panurginus occidentalis (Crawford) second 0.995 Panurgus calcaratus (Spinola) first 0.998 Perdita (Callomacrotera) acapulconis Timberlake second 0.975 Perdita (Perdita) chihuahua Timberlake second 0.999 Protandrena (Heterosarus) neomexicana (Cockerell) first 0.999 Pseudopanurgus aethiops (Cresson) first 0.999 marginata (Cresson) first 0.999 Halictidae altadenae (Michener) first 0.967 Halictinae Lasioglossum (Hemihalictus) lustrans (Cockerell) first 0.999 Dasypoda Panzeri Spinola second 0.999 Dasypodainae Melittidae Hesperapis pellucida Cockerell second 0.985 Melittinae Macropis europaea Warncke first 0.930 Fideliinae Pararhophites orobinus (Morawitz) second 0.598 Anthidium manicatum (Linnaeus) second 0.972 Anthodioctes gualanense (Cockerell) first 0.673 Ashmeadiella bucconis (Say) first 0.905 Megachilidae Atoposmia (Hexosmia) copelandica (Cockerell) first 0.999 Megachilinae Dioxys productus subruber (Cockerell) second 0.999 Megachile chrysopyga Smith second 0.863 Paranthidium jugatorium perpictum (Cockerell) first 0.658 Trichothurgus dubius (Sichel) second 0.894 Allodape interrupta Vachal first 0.892 Compsomelissa (Compsomelissa) nigrinervis (Cameron) second 0.999 Xylocopinae Macrogalea candida Smith second 0.999 Nasutapis straussorum Michener second 0.816 Biastes brevicornis (Panzer) first 0.999 Caenoprosopis crabronina Holmberg second 0.999 Apidae Holcopasites heliopsis (Rober tson) second 0.999 Rhopalolemma Robertsi Roig-Alsina second 0.999 Townsendiella californica Michener second 0.999 Anthophorula (Anthophorula) compactula Cockerell first 0.999 Apinae Ctenoplectra sp. Kirby first 0.999 Eucera chrysopyga Pérez first 0.997 134 Forewings of bees with three submarginal cells (A-H) and two submarginal cells (I-P); the images are not drawn to same scale (redrawn from cells (A-H) and two submarginal F ig. 1. Forewings of bees with three submarginal Michener, 2000). Vol. 55 No. 2 2011 Journal of Apicultural Science 135

Isepeolus viperinus (Holmberg), gibbus (Linnaeus), Stenotritus Lasioglossum (Lasioglossum) leucozonium pubescens (Smith), (Schrank), Leiopodus lacertinus Smith, curvicornis (Scopoli), Tetrapedia sp. Leioproctus (Protomorpha) tarsalis Klug, Thygater analis (Lepeletier), (Rayment), Lipotriches (Austronomia) Trichocolletes venustus (Smith), australica (Smith), Lonchopria Trigonopedia sp. Moure, Xeromelecta zonalis (Reed), Megalopta genalis (Melectomorpha) californica (Cresson), Meade-Waldo, Megandrena enceliae Xylocopa tabaniformis orpifex Smith, (Cockerell), Meliltturga clavicornis and Zacosmia maculata (Cresson). (Latreille), Melissodes agilis Cresson, The wing images (Fig. 1) were Melissoptila (Ptilomelissa) sp. Moure, obtained from the comprehensive book Melitta leporina (Panzer), Mesocheira of Michener (2000). The images were bicolor (Fabricius), Mesonychium scanned using the HP Scanjet 5590 flatbed garleppi (Schrottky), Microshecodes scanner with a resolution of 600 dots per truncaticaudus Michener, Mydrosoma inch. Venation of the wings was compared bohartorum Michener, Neofidelia using landmarks located at wing vein profuga Moure and Michener, junctions (Gerula et al., 2009; Szymula Nomada annulata Smith, Nomia et al., 2010). In wings with two and three (Acunomia) melanderi Cockerell, submarginal cells, coordinates of 16 and Nomioides minutissimus (Rossi), 18 vein junctions (Fig. 2), respectively, Odyneropsis sp. Schrottky, Osiris were determined using tpsDig software sp. Smith, Paranomada velutina (Rohlf, 2005). The coordinates were Linsley, Paratetrapedia calcarata superimposed using the Procrustes (Cresson), Parepeolus niger Roig- method (Rohlf and Slice, 1990). The Alsina, Psaenythia bergi Holmberg, superposition was based on vein junctions Pseudagapostemon sp. Schrottky, 1-14, the homology of which is known. Ptiloglossa guinnae Roberts, Ptilothrix However, in each wing, all 18 (or 16, in fructifer (Holmberg), Sphecodes the case of wings with two submarginal

Fig. 2. Forewings of bees with three submarginal cells (A) and two submarginal cells (B). The vein junctions are numbered with consecutive numbers. Letters indicate submarginal crossveins: f - first submarginal crossvein, s - second submarginal crossvein, r - remaining submarginal crossvein. 136 cells) were translated, scaled, and rotated in In order to compare the size of bees the same way. By doing this, the expected with two and three submarginal cells, the position of the submarginal crossveins midrange of body length was used. The was determined in relation to the rest of midrange was calculated as the sum of the the venation (Fig. 3). The wing diagrams minimum and maximum values, divided presented in Fig. 1 and 2 were produced by two. The minimum and maximum using DrawWing version 0.12 (Tofilski, values for most genera and subgenera 2004). Discriminant function analysis were obtained from Michener (2000). was used to classify the crossveins as the In two cases (Andrena (Callandrena) and first or second submarginal crossvein. Epeolus), the range was not provided in The discriminant analysis was based on the book and the two cases were excluded four variables: two coordinates of anterior from the analysis of size. endpoint and two coordinates of posterior endpoint of crossveins.

Fig. 3. Superimposed vein junctions of bee forewing (A) and enlarged fragment showing only junctions 15-20 corresponding to submarginal crossveins (B). The numbers of the junctions are the same as in Fig. 2. Ellipses indicate area in which the vein junctions can be expected with a probability 0.95. Vol. 55 No. 2 2011 Journal of Apicultural Science 137

Results australica wing (Fig. 1B) the second After superposition most of the vein crossvein was classified as the first and junctions of bee wings formed well in Doeringiella (Triepeolus) verbesinae separated clusters (Fig. 3A). The clusters wing (Fig. 1C), the first crossvein was corresponding to different vein junctions classified as the second. Discrimination overlapped to some degree if distance functions obtained from the discriminant between the junctions was small. This is analysis were used to classify the particularly visible in junctions 12 and 13, remaining crossveins from wings with two which in evolution markedly changed their submarginal cells. Out of 47 remaining position (Fig. 2). crossveins, 22 (46.8%) and 25 (53.2%) In wings with three submarginal cells, were classified as the first and second the first and second crossveins were crossvein, respectively (Tab. 1). relatively well separated, especially, at the Bee genera and subgenera with two posterior ends (junctions 16 and 18; Fig. submarginal cells were significantly 3B, 4AC). The coordinates of the endpoints smaller than bee genera and subgenera differed significantly between the first and with three submarginal cells (Student's second crossveins (MANOVA: Wilks t-test: t=5.62; N1=70; N2=47; P<0.001). Lambda=0.126; F=240.1; P<0.001). The average body length of bees with two In univariate comparisons only the and three submarginal cells was 7.7 and y-coordinate of anterior ends (junctions 11.2 mm, respectively. 15 and 17) did not differ significantly between the two crossveins. Discussion In wings with two submarginal cells, The results clearly show that position the anterior endpoint of the remaining of the remaining submarginal crossveins crossvein formed a single cluster rather than in bees wings with two submarginal cells the two expected clusters. The distribution is markedly different than the position of of the x-coordinate of the anterior endpoint either the first or the second crossveins in (junction 19) of the remaining crossvein wings with three submarginal cells. The was unimodal (Fig. 4B). The posterior endpoints of the remaining crossveins do endpoint (junction 20) of the remaining not form separate clusters corresponding to crossvein also formed a single cluster, either the first or the second submarginal however, in this case the distribution crossvein, as it was expected. The second of x-coordinate was bimodal. The two crossvein moved basad if the first was maxima of the distribution (Fig. 4D) lost and the first crossvein moved apicad did not match corresponding maxima of if the second was lost. In this situation, either the first or the second crossvein reconstruction of homology of the (Fig. 4C). The combined distribution of remaining crossveins is difficult and the x-coordinate of the first and second imprecise. The presented here classification crossvein was significantly different of the remaining crossveins as homologues from the distribution of the x-coordinate with the first or second submarginal of the remaining crossvein in the case of crossvein should be interpreted carefully. both the anterior and posterior endpoint Not only in Hyleoides is it not clear which (Chi-square: χ2=76.2, P<0.001; χ2=146.6, crossvein is missing (Michener, 2000) P<0.001, respectively). but also in many other genera (Tab. 1). Discriminant analysis based on The distribution of the x-coordinate of the resubstitution correctly classified as the posterior end of the remaining crossvein is first or the second crossvein most (98.6%) bimodal (Fig. 4D). This confirms the earlier crossveins of wings with three submarginal suggestions that both the first and second cells. Out of 72 wings with three crossvein can be lost during evolution submarginal cells, two were classified (Peters, 1969). There are two possible incorrectly. In Lipotriches (Austronomia) evolutionary scenarios leading to the 138

Fig. 4. Distribution of x-coordinates of endpoints of submarginal crossveins. Graphs A and B correspond to anterior ends of the crossveins; graphs C and D correspond to posterior ends of the crossveins; graphs A and C correspond to wings with three submarginal cells and graphs B and D correspond to wings with two submarginal cells. Vol. 55 No. 2 2011 Journal of Apicultural Science 139 current position of the remaining crossvein. (e.g. Brachyhesma, Euryglossina, The first scenario is that earlier in evolution, Euryglossula, Neolarra) tend to be smaller one of the crossveins was lost leading to than bees with full venation. It is well one large and one small submarginal cell. known that smaller have reduced Later, the remaining crossvein moved in venation (Peters, 1969; Chapman, such a way that the disproportion of cell 1998). The reduction can range from loss sizes was reduced. The second scenario of distal veins (Danforth, 1989) to loss of is that earlier in evolution, two crossveins almost all veins, as in Chalcidoidea (Gauld moved in such a way that there were three and Bolton, 1988). submarginal cells of unequal sizes and later, one of the crossveins was lost leading Conclusions to two submarginal cells of similar sizes 1. The data presented here suggest (Robertson, 1926). Disappearance of that in bees there are patterns of wing the crossveins can either be sudden, as it venation which are preferred by natural sporadically happens in Halictus (Peters, selection. Similar patterns can be a result 1969), or gradual like in Lasioglossum of the disappearance of either the first (Michener, 2000). or the second submarginal crossvein. In In some studies, it was assumed this situation, it is difficult to determine that differences in wing venation are homology of the veins using their position proportional to phylogenetic differences alone. and they were used for reconstructions 2. Genera with three submarginal cells of phylogenies (Dietrich et al., 2001). tend to be larger than bees with two The results presented here suggest that in submarginal cells. bees, the positions of the crossveins are not independent of each other and natural Acknowledgements selection favours some patterns of veins. I wish to thank Bernadetta Rzeźnicka Therefore, wings with similar venation for her help digitizing the landmarks. This can be a result of convergent evolution work was supported by MNiSW grant No. (Sharkey and Roy, 2002). Shape of N N311 292436. venation can affect fitness because veins are important for wing stiffness and in References consequence, for flight performance (Combes and Daniel, 2003). On the other Alpatov W.W. (1928) - Variation of hand, there is evidence that in honeybees hooks on the hind wing of the honey bee (Apis mellifera), variation of wing venation (Apis mellifera L.). Biol. Bull., 55: 209-234. present in natural populations does not Chapman R. F. (1998) - The insects: structure affect fitness (Diniz-Filho et al., 1999). and function. Cambridge University Press, The loss of one of the crossveins can be Cambridge, 788 pp. related to size. Bees with three submarginal Combes S. A., Daniel T. L. (2003) cells are on average, larger than bees with - Flexural stiffness in wings. two submarginal cells. Comparison of I. Scaling and the influence of wing venation. subgenera within some genera confirms this. J. Exp. Biol., 206: 2979-2987. The genus Eucera subgenus Synhalonia with the largest bees, has three submarginal Danforth B. N. (1989) - The evolution of cells, and the other subgenera with smaller hymenopteran wings: the importance of size. bees have two submarginal cells. Genus J. Zool., 218: 247-276. Leioproctus subgenus Filiglossa with Dietrich C. H., McKamey S. H., the smallest bees has two submarginal Deitz L. L. (2001) - Morphology- cells while most other subgenera with based phylogeny of the treehopper family larger bees have three submarginal cells. Membracidae (Hemiptera: Cicadomorpha: Moreover, bees with incomplete venation Membracoidea). Sys. Entomol., 26: 213. 140

Diniz-Filho J. A. F., Fuchs S., Arias M. C. Robertson C. (1926) - Wing veins of bees as (1999) - Phylogeographical autocorrelation strengthening elements. Psyche, 33: 39-41. of phenotypic evolution in honey bees Rohlf F. J., Slice D. (1990) - Extension (Apis mellifera L.). Heredity, 83: 671-680. of the Procrustes method for the optimal Gauld I., Bolton B. (1988) - The Hymenoptera. superimposition of landmarks. Syst. Zool., 39: Oxford University Press, Oxford, 332 pp. 40-59. Gerula D., Tofilski A., Rohlf F. J. (2005) - tpsDig, digitize landmarks Węgrzynowicz P., Skowronek W. and outlines, version 2.04. Department of (2009) - Computer-assisted discrimination Ecology and Evolution, State University of of honeybee subspecies used for breeding in New York at Stony Brook. Poland. J. Apic. Sci., 53(2): 105-114. Sharkey M. J., Roy A. (2002) - Phylogeny of Louis J. (1973) - La nomenclature de l’aile des the Hymenoptera: a reanalysis of the Ronquist hyménopte`res, essai de normalisation. Beitr. et al. (1999) reanalysis, emphasizing wing Entomol., 23: 275-289. venation and apocritan relationships. Michener C. D. (2000) - The bees of the world. Zool. Scripta, 31: 57-66. Johns Hopkins University Press, Baltimore, Szymula J., Skowronek W., 913 pp. Bieńkowska M. (2010) - Use of various Peters D. S. (1969) - Phänokopierende morphological traits measured by microscope Mißbildungen des Flügelgeäders bei akuleaten or by computer methods in the honeybee Hymenopteren. Deut. Entomol. Z., 16: 367- . J. Apic. Sci., 54(2): 91-97. 374. Tofilski A. (2004) - DrawWing, a program Robertson C. (1922) - Synopsis of Panurgidae for numerical description of insect wings. (Hymenoptera). Psyche, 29: 159-173. J. Insect Sci., 4: 1-5. Homologia żyłek submarginalnych przednich skrzydeł pszczół (Hymenoptera: Apiformes)

Tofilski A .

Streszczenie Na przednich skrzydłach pszczół (Apiformes) znajdują się dwie lub trzy komórki submarginalne. W przypadku skrzydeł z dwoma komórkami submarginalnymi nie wiadomo czy żyłka oddzielająca te dwie komórki (zwana dalej żyłką problematyczną) jest homologiczna z pierwszą czy drugą żyłką submarginalną. Poznanie homologii tej żyłki jest ważne w przypadku odtwarzania filogenezy pszczół. Podjęto próbę określenia homologii problematycznej żyłki używając metod ilościowych. Współrzędne 14 połączeń żyłek przedniego skrzydła nałożono na siebie w celu określenia położenia żyłek oddzielających komórki submarginalne. Oczekiwano, że rozkład położenia problematycznej żyłki będzie podobny do połączonych rozkładów położenia pierwszej i drugiej żyłki submarginalnej. Wbrew oczekiwaniom okazało się, że problematyczna żyłka znajduje się często pomiędzy oczekiwanym położeniem pierwszej i drugiej żyłki submarginalnej. Rozkład położenia problematycznej żyłki był dwumodalny, co potwierdza wcześniejsze sugestie, że zarówno pierwsza jak i druga żyłka submarginalna może ulec zanikowi w czasie ewolucji. Uzyskane wyniki wskazują, że u pszczół istnieje wzór użyłkowania, który jest preferowany przez dobór naturalny. Podobny wzór użyłkowania może się pojawić w wyniku zaniku zarówno pierwszej jak i drugiej żyłki submarginalnej. Utrudnia to ustalenie homologii problematycznej żyłki. Słowa kluczowe: pszczoły, Apiformes, skrzydło, użyłkowanie, komórki submarginalne, homologia.