Original article

A morphological comparison of the dwarf honey bees of southeastern and Palawan,

TE Rinderer BP Oldroyd S Wongsiri HA Sylvester LI de Guzman JA Stelzer RM Riggio

1 USDA, Agricultural Research Service, Honey-Bee Breeding, Genetics and Physiology Research Laboratory, 1157, Ben Hur Rd, Baton Rouge, LA 70820 USA; 2 Bee Biology Research Unit, Department of Biology, Chulalongkorn University, Bangkok 10330, Thailand; 3 Dept of Entomology, Louisiana State University, Baton Rouge, LA 70893, USA

(Received 20 February 1995; accepted 11 April 1995)

Summary — Morphological descriptions using measurements common to are pro- vided for Apis andreniformis Smith (1858) from southeastern Thailand and Palawan, Philippines and Fabricius (1787) from southeastern Thailand. Overall, A andreniformis has a very different morphology from the sympatric A florea and from all other well-described species of honey bees. One difference, the color of the scutellum, although not species specific, is sufficient in many cases to facilitate field identifications. Very few morphological differences occurred between the Thai and the Philip- pine populations of A andreniformis.

Asian bees / Apis andreniformis / Apis florea / morphometry / population genetics / Thailand / Philippines

INTRODUCTION study supported the conclusions of Wu and Kuang (1986, 1987) who studied furcated The small dwarf honey bee, Apis andreni- basitarsus differences between drones of formis Smith (1858) has recently been re- A florea Fabricius (1787) and A andreni- evaluated and recognized as a valid bio- formis. A species-specific characteristic of A logical species based on morphological andreniformis identified by Smith (1858) of evidence of a unique endophallus, charac- worker bees having black hairs on the hind teristic worker bee wing venation, and a dis- tibia and dorsolateral surface of the hind tinctive furcation of the male basitarsus basitarsus as opposed to the white hairs of (Wongsiri et al, 1990). The Wongsiri et al A florea was confirmed by these studies. Further confirmation of the valid biological the 2 species are sympatric. Four colonies of A andreniformis were from species status of A andreniformis was pro- sampled Palawan, Philip- an island which is outside the distribution vided by Rinderer et al (1993a) who reported pines, of A florea. Ten workers from each colony were that the of drones from mating flights sym- dissected and 44 morphometric measurements patric A andreniformis and A florea were were made using computer-assisted techniques temporally separate. While these reports (Daly et al, 1982; Rinderer et al, 1993b). These clearly show A andreniformis to be a sepa- measurements and their alphabetical designa- rate species, they do not describe all the tions are given in table I and follow the system of described Ruttner morphological measurements commonly morphometric analysis by et al (1978), Ruttner (1987) and Rinderer et al used in modern bee honey taxonomy (Rut- (1989). Numerical designations for wing vena- This mor- tner, 1987). paper provides these tion angles (fig 1) are a combination of the angles phometric details for 2 populations of A studied by Ruttner et al (1978) and Daly and andreniformis occurring in southeastern Balling (1978) and follow the designations of Thailand and Palawan, Philippines (de Guz- Rinderer et al (1989). For the most part, num- bers represent lengths or widths of various struc- man et al, 1992). tures and are reported in millimeters. Interior Since A andreniformis is most similar to angles of vein intersections in wing venation pat- A florea and since little is known of the inter- terns are reported in degrees. Pigmentation char- species morphological variation of either acteristics of the second, third and fourth tergites, species, comparative data for the sympatric the scutellum, the metanotum and the mesonotum are scored according to the procedures of Ruttner of A florea in Thailand are also population et al (1978) and Ruttner (1987). Scores for ter- provided. gite color intensity range from 0 to 10 with 0 being completely black and 10 being completely yel- low. Scutellum scores range from 0 to 9 with 0 MATERIALS AND METHODS being completely black and 9 being completely yellow. Scores for the color intensity of the metan- otum (B) and the mesonotum (K) range from 0 Twenty-seven colonies of A florea and 18 colonies to 5 with completely black being 0 and completely A andreniformis were sampled in Thailand where yellow being 5.

After measurements were made on individ- The 2 populations of A andreniformis dif- ual bees, colony averages for each characteristic fered for only a few characters. The Thai were calculated. These averages were used to A andreniformis had and calculate variances and for each longer forewings means, ranges but fewer hamuli. measurement within each species (table I). The longer hindwings, Wing differences with similar widths colony averages were also used to calculate t- length wing test evaluations of differences between A florea probably led to the 3 observed significant and A andreniformis from Thailand and between differences in wing angles. Thai A andreni- A andreniformis from Thailand and the Philip- formis also had significantly narrower A multivariate discriminant was pines. analysis metatarsi and a scutellum which was, on also conducted to estimate Mahalanobis dis- than that of both A florea tances between population centroids. average, lighter and A andreniformis from the Philippines. A multivariate discriminant analysis of RESULTS data summarizes the population differences. A Mahalanobis difference of 110 252 sepa- rates Thai A andreniformis from Thai A flo- In general, A florea from Thailand are larger rea while a Mahalanobis difference of 2 891 than sympatric A andreniformis. Of the 17 separates Thai A andreniformis from Philip- measurements of length or width involving pine A andreniformis (fig 2). wings, legs, mouthparts and sternites, in only 2 cases was A andreniformis equal to or than A florea. both greater Interestingly, DISCUSSION these cases involved wing length: forewing lengths were about equal and hindwing lengths of A andreniformis were larger. Overall, A andreniformis has a very different The 2 species differed in all the angles of morphology from the sympatric A florea. wing vein intersection save number 40. The color of the scutellum appears to be a Cubital segment A was larger in A andreni- useful field character to identifying A formis and cubital segment B was larger in andreniformis. Additionally, the remainder A florea. Consequently, the cubital index of of the measurements provide a picture of A andreniformis (6.37) was much larger than A andreniformis as a remarkably different that of A florea (2.86). honey bee, unique in its morphological orga- nization. In general, A florea had less black pig- ment on the various structures measured The differences between the Thai and in congruence with the general visual the Philippine populations of A andreniformis impression that A florea are mostly yellow are rather small considering their geo- bees and A andreniformis are mostly black graphical separation. Similar distances with bees. A notable exception to this rule is the geographical isolation result in strong sub- pigmentation of the scutellum. Scutellum specific variation in A mellifera (Ruttner, color for A andreniformis tends toward yel- 1987). However, the differences between low while the scutellum color for A florea the Thai and Philippine populations of tends toward black. The range of A andreni- A andreniformis are less than differences formis scutellum coloration does extend into typically used to support subspecific desig- the black range. However, the range of nations. But the 2 groups show no overlap in scutellum coloration for A florea of our sam- the multivariate discriminant analysis pro- ple does not include yellow. Thus, this highly cedures with each group forming a tight visible character may be useful for field iden- grouping about its centroid. If more sam- tifications of A andreniformis. ples confirmed these observations, the lack of overlap might be interpreted as indicating development of honey bee characteristics subspecific differences (Ruttner, 1987). (Rinderer et al, 1992, 1993a). Much remains to be learned about A andreniformis. Its historical confusion with ACKNOWLEDGMENT A florea has probably resulted in many pub- lished studies on what was actually A andreniformis being reported as informa- This study was made in cooperation with tion on A florea. Clear information on the Louisiana Agricultural Experiment Station. ranges of these 2 species will provide a guide as to which literature resulted from studies of bees in areas of sympatry and Résumé — Comparaison morphologique needs to be considered suspect. The dwarf des abeilles naines du sud-est de la Thaï- honey bees represent great potential for lande et de l’ile de Palaouan (Philippines). comparative studies in honey bee biology La petite abeille naine, Apis andreniformis and, indeed, have already provided some Smith (1858), a été récemment confirmée interesting insight into the evolutionary comme étant une véritable espèce biolo- gique (Wongsiri et al, 1990). Bien que cette a moins de pigment noir, ce qui est publication, ainsi que d’autres (Wu et Kuang, conforme à l’impression générale qu’A florea 1986, 1987; Rinderer et al, 1993a), le est principalement une abeille jaune alors démontrent clairement, ces travaux étaient qu’A andreniformis est principalement une incomplets puisqu’ils n’incluaient pas toutes abeille noire. les mesures morphologiques habituellement Les différences entre les populations d’A utilisées en taxonomie moderne (Ruttner, andreniformis de Thaïlande et des Philip- 1987). Ces mesures sont présentées dans pines sont plutôt petites malgré leur sépa- le présent travail, elles concernent 2 popu- ration géographique. Cependant, les 2 lations d’A andreniformis du sud-est de la groupes ne présentent pas de recouvre- Thailande et de l’ile de Palaouan (Philip- ment dans l’analyse discriminante multiva- Guzman pines) (de et al, 1992). riée, et chaque groupe forme un groupe- Vingt-sept colonies d’A florea et 18 colo- ment serré autour de son centroïde. Le nies d’A andreniformis ont été échantillo- manque de recouvrement peut être inter- nées en Thaïlande où les 2 colonies sont prété comme indiquant des différences sympatriques. Quatre colonies d’A andre- sous-spécifiques (Ruttner, 1987). niformis ont été échantillonnées à Palaouan, Par dessus tout, A andreniformis pré- une île est florea. qui allopatrique pour A sente une morphologie très différente d’A Dix ouvrières de chaque colonie ont été dis- florea sympatrique. La couleur du scutellum et 44 mesures séquées morphométriques apparaît être un caractère très utile dans la ont été réalisées en utilisant une méthode de nature pour identifier cette espèce. De plus, mesures assistée ordinateur et par (Daly l’ensemble des mesures montre qu’A andre- Rinderer Ces al, 1982 ; et al, 1993b). niformis est une espèce d’abeille remar- mesures et leurs désignations alphabétiques quablement différente des autres, et unique dans le tableau I suivent la présentées dans son organisation morphologique. méthode d’analyses morphométriques décrite par Ruttner et al (1978), Ruttner abeille asiatique / Apis andreniformis / (1987) et Rinderer et al (1989). Après que Apis florea / morphométrie / génétique les mesures ont été réalisées sur les abeilles des populations / Thaïlande / Philippines individuelles, les moyennes des colonies pour chaque caractéristique ont été calcu- lées, ainsi que les moyennes, les variances Zusammenfassung — Ein morphologi- et les écarts pour chaque espèce (tableau I). scher Vergleich der Zwerghonigbienen Les colonie ont moyennes par également von Südost-Thailand und Palawan, Phil- été utilisées pour réaliser le test t sur les lipinen. Die Zwerghonigbiene, Apis andre- différences entre les colonies d’A florea et niformis Smith (1858) wurde erst kürzlich d’A andreniformis de Thaïlande et entre les erneut kritisch untersucht und als eigene colonies d’A andreniformis des Philippines. biologische Art anerkannt (Wongsiri et al, En général, A florea est, en Thaïlande, 1990). Diese und andere Arbeiten über A plus grande qu’A andreniformis. Ces 2 andreniformis (Wu and Kuang, 1986, 1987; espèces diffèrent dans tous les angles d’in- Rinderer et al, 1993a) zeigen deutlich, daß tersection des veines des ailes, sauf une. es sich wirklich um eine Art handelt. Sie sind Le segment cubital A est plus grand chez dennoch unvollständig, da sie nicht alle mor- A andreniformis et le segment cubital B chez phologischen Eigenschaften beschreiben, A florea. Par conséquent, l’index cubital d’A die in der modernen Taxonomie der Honig- andreniformis (6,37) est beaucoup plus bienen im allgemeinen angegeben werden grand que celui d’A florea (2,86). A florea (Ruttner, 1987). Die jetzige Veröffentlichung beschreibt diese morphologischen Einzel- tendiert bei A andreniformis mit einigen Aus- heiten für 2 Populationen von A andrenifor- nahmen zum gelblichen, während es bei A mis aus Südost-Thailand und Palawan von florea zu dunkler Farbe tendiert. den Philippinen (de Guzman et al, 1993). Die Unterschiede zwischen der Thai- und Siebenundzwanzig Völker von Apis florea der philippinischen Population sind recht Fabricius (1787) und 18 A andreniformis gering, wenn man die geographische Ent- Völker wurden in Thailand gesammelt, wo fernung bedenkt. Beide Gruppen zeigen beide Arten sympatrisch vorkommen. Vier A jedoch keine Überlappung bei einer multi- andreniformis Völker stammen von Pala- variaten Diskriminanzanalyse. Jede Gruppe wan, einer Insel auf der es keine A florea bildet eine enge Gruppierung um ein Cen- gibt. Von jedem Volk wurden 10 Arbeiterin- troid. Das Fehlen einer Überlappung kann nen präpariert und 44 morphometrische als rassenspezifische Unterschiede gedeu- Daten mit einer Computer unterstützten tet werden (Ruttner, 1987). Technik (Daly et al, 1982; Rinderer et al, Insgesamt sind A andreniformis und die erhoben. Diese Meßdaten und ihre 1993b) sympatrische A florea morphologisch sehr alphabetische Einordnung sind in Tabelle I unterschiedlich. Die Farbe des Scutellums die auf dem der wiedergegeben, System scheint eine nützliche Eigenschaft für eine morphometrischen von Ruttner et al Analyse Feldbestimmung zu sein. Die übrigen Mes- Ruttner und Rinderer et al (1978), (1987) sungen zeichnen das Bild einer erstaunlich beruhen. (1989) unterschiedlichen Honigbienenart, die in Nach Messungen der Einzelbienen ihrer morphologischen Organisation ein- wurde der Volksdurchschnitt für jede Eigen- zigartig ist. schaft berechnet. Diese Durchschnittswerte wurden benutzt, um Mittelwerte, Varianzen Asiatische Bienen / Apis andreniformis / und Bereich für jeden Meßwert innerhalb Apis florea / Morphometrie / Populati- jeder Art zu berechnen (Tabelle I). Die onsgenetik / Thailand / Philippinen Unterschiede der Volksmittelwerte zwischen A florea und A andreniformis in Thailand und zwischen A andreniformis von Thailand REFERENCES und von den Philippinen wurden mit dem t- Test bestimmt. Daly HV, Balling SS (1978) Identification of Africanized A florea in Thailand ist insgesamt größer honey bees in the western hemisphere by discrimi- als die sympatrische A andreniformis. Die nant analysis. J Kans Entomol Soc 51, 857-869 beiden Arten unterschieden sich in allen Daly HV, Hoelmer K, Norman P, Allen T (1982) Com- measurement and identification of Winkeln des bis auf einen. puter-assisted Flügelgeäders honey bees (: ). Ann Entomol Das Cubitalsegment A war größer in A an- Soc 75, 591-594 dreniformis und das Cubitalsegment B war Fabricius JC (1787) Mantissa Insectorum. Vol I Proft, größer in A florea. Entsprechend war der Hafniae Cubitalindex von A andreniformis mit 6,37 Guzman de LI, Forbes M, Cervancia C, Rinderer TE, viel größer als der von A florea mit 2,86. Somera S Jr (1992) Apis andreniformis Smith in Palawan, Philippines. J Apic Res 31, 111 A florea hat schwarzes weniger Pigment. Rinderer TE, Koeniger N, Tingek S, Mardan M, Koeniger Das stimmt mit dem Eindruck überein, daß G (1989) A morphological comparison of the cavity es sich bei A florea um überwiegend gelb- dwelling honeybees of Borneo Apis koschevnikovi and cerana liche und bei A andreniformis um überwie- (Buttel-Reepen, 1906) Apis (Fabricius, 1793). Apidologie 20, 405-411 gend dunkle Bienen handelt. Eine auffällige Rinderer TE, Oldroyd BP, Wongsiri S, Guzman de LI, Ausnahme zu dieser ist die Regel Färbung Sylvester HA (1992) Evolution of bee dances. Nature des Scutellums. Die Farbe des Scuttellums (Lond) 306, 305 Rinderer TE, Oldroyd BP, Wongsiri S et al (1993a) Time Smith F (1858) Catalogue of the Hymenopterous of drone flight in 4 honey bee species in southeast- collected at Sarawak, Borneo, Mount Ophir, Malacca, ern Thailand. J Apic Res 32, 27-33 and at Singapore, by AR Wallace. J Prov Linn Soc Rinderer TE, Buco SM, Rubink WL et al (1993b) Mor- Lond 2, 42-130 phometric identification of Africanized and European Wongsiri S, Limbipichai K, Tangkanasing P et al (1990) honey bees using large reference populations. Api- Evidence of reproductive isolation confirms that Apis dologie 24, 569-585 andreniformis Smith (1858) is a separate species Ruttner F (1987) Biogeography and Taxonomy of the from sympatric Apis florea (Fabricius, 1787). Api- Honeybees. Springer Verlag, Berlin, 284 p dologie 21, 47-52 Ruttner F, Tassencourt L, Louveaux J (1978) Biometri- Wu Y, Kuang B (1986) A study of the Micrapis cal-statistical analysis of the geographic variability (Apidae). Zool Res 7, 99-102; In: Chinese Bee World of Apis mellifera L. Apidologie 9, 363-381 68, 153-155 (a translation of Zool Res 7, 99-10)