Multivariate analysis of honeybees, Apis mellifera Linnaeus (Hymenoptera: Apidae) of the northeastern and southern regions of Algeria C. Barour 1 , A. Tahar 1 , S.E. Radloff 2 * & H.R. Hepburn 3 1 Departement de Biologie, Laboratoire de Biologie Animale Appliquée, Université Badji Mokhtar, Annaba, BP 12 Algeria 2 Department of Statistics, Rhodes University, Grahamstown, 6140 South Africa 3 Department of Zoology and Entomology, Rhodes University, Grahamstown, 6140 South Africa Morphometric characters of worker honeybees, Apis mellifera Linnaeus, were analysed by multivariate methods to characterize their populations in the northeastern and southern regions (ecological-climatological zones) of Algeria, from 36°55’ to 32°25’N and 3°39’ to 8°22’E. Three morphoclusters were identified by using principal components and linear discriminant analyses. The bees are identified as Apis mellifera intermissa Buttel-Reepen. Key words: Apis mellifera intermissa, morphometrics, ecological-climatological zones, Algeria. INTRODUCTION Previous analyses of the honeybees of north- western Africa established that Apis mellifera comprised several different populations. Those of southern Morocco formed a discrete morpho- cluster, the subspecies A. m. sahariensis, while those of northern Morocco, Tunisia and northwestern Algeria constituted differentiated populations of A. m. intermissa (Grissa et al. 1990; Cornuet et al. 1988; Hepburn & Radloff 1996). Nonetheless, the apifauna of northeastern Algeria and Libya have not yet been analysed morphometrically or taxo- nomically. Recent collections of honeybees from the eastern section of Algeria have been made that fill a former lacuna between Morocco and Tunisia, thus providing data along a contiguous 2000 km transect along Mediterranean North Africa. Here we report on the results of multivariate morpho- metric analyses of these bees and consider their relationship to other honeybee populations of the northern Maghreb, i.e. Morocco, Algeria and Tunisia. MATERIAL AND METHODS Samples of worker honeybees were collected from 39 colonies of Apis mellifera at 12 different localities in seven provinces from different geo- graphic regions, situated between latitudes 36°55’ and 32°25’N and longitudes 3°39’ and 8°22’E, in the north-east and the south of Algeria and were used in the analysis (Table 1, Fig. 1). Colonies sampled were mostly taken from fixed-site beekeepers. Samples were preserved in 90 % ethanol and bees were dissected according to the methods given by Ruttner (1988). Morphometric measure- ments were usually made on 20 bees per colony (except the locality of Serraïdi where 12 bees per colony were analysed) and a total of 756 bees was used in the morphometric analysis. Nineteen morphometric characters, all based on the Oberursel standard list (Ruttner 1988), were measured; they were: forewing length [01], forewing width [02], cubital vein A length [03], cubital vein B length [04], hindwing length [05], hindwing width [06], hamuli number [07], femur length [08], tibia length [09], metatarsus length [10], metatarsus width [11], length sternite 2 [12], length sternite 3 [13], length of wax mirror on sternite 3 [14], width of wax mirror on sternite 3 [15], distance between wax mirrors on sternite 3 [16], pigmentation of abdominal tergite 2 [17], width of tomentum on tergite 4 [18], length of cover hair on tergite 5 [19]. Multivariate statistical analysis of the data in- cluded principal components, point density, k-means cluster and linear discriminant analyses. Wilks’ lambda test was used to compare multi- variate population means between groups. The distribution of the statistic was approximated by the F-distribution (Johnson & Wichern 1998). Levene’s statistic for testing the equality of inter- and intra-colonial variances between localities was also used in the analysis. RESULTS In a principal components analysis of the stan- dardized morphometric characters of worker honeybees from 39 colonies, five factors (i.e. African Entomology 13(1): 17–23 (2005) *To whom correspondence should be addressed, E-mail: s.radloff@ru.ac.za