Printer Friendly

The origin of west European subspecies of honeybees (Apis mellifera): new insights from microsatellite and mitochondrial data.

Before they were disseminated around the world by humans, populations of the western honeybee, Apis mellifera, were naturally distributed in Africa, Europe, and western Asia. In this vast area, they have profoundly differentiated; 24 subspecies have been recognized through multivariate analysis of morphological characters (Ruttner et al. 1978; Ruttner 1988; [ILLUSTRATION FOR FIGURE 1A OMITTED]). On the same morphometric basis, these subspecies have been grouped in three main evolutionary branches including the African subspecies (branch A), northern Mediterranean subspecies (branch C) and western European subspecies (branch M).

Mitochondrial DNA (mtDNA) studies have shown that haplotypes also fall into three groups, each one associated with a previously defined branch. The agreement with morphometric data is so close that A, C, and M have been used to designate the three families of mtDNA haplotypes (Smith 1991; Garnery et al. 1992). The same general structure of the species emerged from a microsatellite survey of a sample of populations taken from the three evolutionary branches (Estoup et al. 1995a). The overall agreement achieved with three different sets of markers suggests that these branches were separated a long time ago and that current subspecies have differentiated within each branch in an area close to their present distribution.

Although morphometric, mtDNA, and microsatellite studies agree on a large scale about the evolutionary history of the species, there is still an important point on which researchers come to opposite conclusions. This concerns the origin of the western European populations of branch M, which is composed of only two subspecies, A.m. mellifera and A.m. iberica. These populations have been extensively studied through morphometry (Cornuet et al. 1975, 1978, 1982, 1988; Ruttner 1988; Cornuet and Fresnaye 1989; Orantes-Bermejo and Garcia-Fernandez 1995; Hepburn and Radloff 1996), enzyme electrophoresis (Cornuet 1982, 1983; Smith and Glenn 1995), and mtDNA (Smith et al. 1991; Garnery et al. 1992, 1993, 1995).

According to morphometric data, colonization of western Europe by honeybees would have occurred from North Africa across the Strait of Gibraltar. In fact, there is a clear morphological gradient extending from the subspecies intermissa in Mahgreb to the subspecies mellifera in western Europe, with the Spanish and Portuguese subspecies iberica showing intermediate traits. This gradient was interpreted by Ruttner et al. (1978) as the indication that the Iberian Peninsula was a large area of primary intergradation between A. m. intermissa and A. m. mellifera. This conclusion was supported by a parallel cline at the MDH locus (Cornuet 1983; Smith and Glenn 1995), a cline that extends from Morocco to France through the Iberian Peninsula. However, another polymorphic allozyme marker, PGM, has similar frequencies in the south and north of Spain (Smith and Glenn 1995).

In Spain, mtDNA haplotypes A (African) are found in high frequencies and also form a north-south ratio cline with haplotypes M (western Mediterranean; Smith et al. 1991; Garnery et al. 1995). The two categories of haplotypes A and M are highly divergent (2.6% from restriction data on the whole molecule; Garnery et al. 1992), so that the hypothesis that haplotypes M derive from haplotypes A through accumulation of mutations during a colonization from south to north is untenable. Rather, this high divergence suggests that the Iberian Peninsula is an area of secondary intergradation between two deeply diverged branches (Smith et al. 1991). However, this new hypothesis is supported by only a single type of marker (mtDNA). In addition, the relationship between North African- and Spanish-lineage haplotypes A is less simple than expected if Spanish-lineage haplotypes A had been recently introduced from Morocco: the lineages make two distinct clades in a neighbor-joining tree based on two genetic distances (Garnery et al. 1995).

To evaluate these hypotheses and better understand the genetic origin of the two western European subspecies, A. m. mellifera and A. m. iberica, we surveyed 11 populations along a transect from Guinea to France for both mitochondrial and nuclear markers. The mitochondrial analysis was based on the COI-COII intergenic sequence (Garnery et al. 1993, 1995) and the nuclear markers were microsatellite loci (Estoup et al. 1995a). If the hypothesis of primary intergradation is correct, we expect the genetic variability of populations to show gradual modifications at microsatellite loci along the transect, with a possible reduction northward due to colonization effects dependent on age. If the intergradation is secondary, we should find frequency clines of African and European alleles in Spain that parallel those for mtDNA. However, neither of the two possibilities was observed with microsatellite data, implying the rejection of both previous hypotheses in favor of a new one.


Origin of Samples

Eleven populations belonging to six subspecies and distributed along a transect from northern France to Guinea were sampled [ILLUSTRATION FOR FIGURE 1B OMITTED]. An Italian population sample from Emily-Romagne (A. m. ligustica, C lineage) was used as an outgroup in phylogenetic analyses. MtDNA and microsatellite data included and expanded upon those published by Garnery et al. (1995) and Estoup et al. (1995a). The Spanish samples from the Basque Country and Andalucia were combined with the colonies from San Sebastian and Sevilla, respectively, already described by Garnery et al. (1995). The sample from Castile was actually composed of two subsamples taken in Segovia and Toledo (Garnery et al. 1995) because each sample size was considered too small and their microsatellite allelic frequencies were very similar (Segovia and Toledo are located 125 km apart). In Morocco, the southwest Rif and southern Morocco samples were identical to those given by Garnery et al. (1995). The samples from northern France, Vaucluse, southern Morocco, and Italy included the Valenciennes, Avignon, Tiznit, and Forli samples of Estoup et al. (1995a).

Worker bees were brought to the laboratory in ethanol. Total DNA was extracted from worker thoraces according to Kocher et al.'s (1989) protocol, with slight modifications as described in Garnery et al. (1991). These extracts were used for mitochondrial DNA and microsatellite analyses.

Mitochondrial DNA

The region between the COI and COII genes contains the gene for [tRNA.sup.leu] and a noncoding sequence, the length of which varies between ca. 200 bp and 860 bp. This sequence includes a combination of two components, P and Q (Cornuet and Garnery 1991; Cornuet et al. 1991). The sequence P has two forms, P and Po, which differ by an insertion/deletion (P is 15 bp shorter than Po). The other sequence, Q, is 194196 bp long and its 3[prime] end ([Q.sub.3]) is almost identical to Po. Nine combinations have been observed in this noncoding sequence: Q, PQ, PQQ, PQQQ, PQQQQ, PoQ, PoQQ, PoQQQ, and PoQQQQ Q is characteristic of bees from lineage C. The four combinations with P (PQ to PQQQQ) are characteristic of lineage M, and the latter three with Po (PoQ to PoQQQ) are characteristic of lineage A (Garnery et al. 1992).

A fast test including the PCR amplication of the COI-COII region and the sizing of the PCR product before and after a DraI (TTT[arrow down]AAA) restriction, has been designed by Garnery et al. (1993). This test, coupled with the precise knowledge of many reference sequences, provides the composition in P/Po and Q(s) of the noncoding sequence and hence discriminates the three A, M, and C lineages. Furthermore, this test reveals many haplotypes in lineages A and M due to the superimposition of length, DraI site, and short insertion/deletion polymorphisms.

The COI-COII region was PCR-amplified with primers E2 (5[prime] GGCAGAATAAGTGCATTG3[prime]) and H2 (5[prime]CAATATCATTGATGACC3[prime]) located in the [tRNA.sup.leu] and COII genes, respectively (for PCR conditions, see Garnery et al. 1993), An aliquot of the PCR product was electrophoresed in a 1% agarose gel to determine its size. The remaining amplified DNA was digested with DraI (Boehringer), and restriction fragments were separated on 5% and 10% polyacrylamide gels and UV visualized after staining with ethidium bromide.

Mitochondrial haplotypes were determined for 341 new colonies (one bee per colony) from seven populations that, when added to the 233 already scored (Garnery et al. 1995), made a total of 574 in the present study. Every new haplotype was thoroughly characterized through sequencing, which allows the precise determination of the size of DraI restriction fragments and the detection of small insertions/deletions.

Microsatellite Loci

Eight microsatellite loci were scored: A43, BI24, A88, A113, A28, A24, and A7 (previously described by Estoup et al. 1995a) and A8 (a new locus). Primer sequences for locus A8 were 5[prime]CGAAGGTAAGGTAAATGGAAC3[prime] and 5[prime]GGCGGTTAAAGTTGTGG3[prime]. The annealing temperature was 55 [degrees] C, and Mg[Cl.sub.2] concentration was adjusted to 1.2 mM. PCRs and electrophoresis were conducted as in Estoup et al. (1995 a).

Microsatellites were generally scored on subsamples. Analysis included eight loci scored in 328 individuals: 225 individuals were entirely new and locus A8 was added to the 103 individuals (from northern France, Vaucluse, southern Morocco, and Italy) already scored at the other seven loci (Estoup et al. 1995a).

Statistical Analyses

Unbiased estimates and standard deviations of gene diversity of mtDNA (D) and microsatellite loci (H) were calculated according to Nei and Tajima (1981) and Nei (1978), respectively, Differences in average unbiased diversity between population samples were assessed by Wilcoxon's signed rank test (Shedecor and Cochran 1978). Exact tests for Hardy-Weinberg equilibrium, genotypic linkage disequilibrium, and genetic structure (genic and genotypic differentiation) were computed using GENEPOP (vers. 1.2; Raymond and Rousset 1995). Multiple test significance was assayed using Fisher's exact test method (GENEPOP) and/or applying Bonferroni's procedure. Isolation by distance was tested according to Slatkin (1993), with []-values being estimated and tested with GENEPOP.

To detect recent changes of the population effective sizes, observed gene diversities ([H.sub.c]) were compared to their expected values under the assumption of mutation-drift equilibrium ([H.sub.e]) at each locus x sample combination under the infinite allele (IAM) and the stepwise mutation (SMM) models as described in Cornuet and Luikart (1996).

A phylogeny of mitochondrial haplotypes was reconstructed using PAUP (vers. 3.1.1; Swofford 1989). The presence/absence of restriction sites and small insertions/deletions along the intergenic sequence was coded as 1/0. When a component was missing (e.g., P/Po in lineage C haplotypes [TABULAR DATA FOR TABLE 1 OMITTED] or additional Q(s) in lineages M or A), the corresponding characters were coded as missing data.

Relationships among population samples were established using two character-based methods, maximum parsimony (MP) and maximum likelihood (ML), and a distance-based method, neighbor-joining (NJ; Saitou and Nei 1987). The Wagner MP method and the ML method were computed on microsatellite and mitochondrial data using the mix and contml procedures of PHYLIP (vers. 3.5c; Felsenstein 1993) and/or PAUP. Alleles and haplotypes were coded as presence/absence characters in each population sample. The characters were weighted by allele or haplotype frequencies for the ML algorithm. For mitochondrial data, we used a distance ([D.sub.A], Garnery et al. 1995) that is analogous to the shared allele distance ([D.sub.AS]) defined for nuclear genes (Chakraborty and Jin 1993): the distance between two populations is the average of the distances between two individuals, one taken in one population and the other taken in the other population. The distance between two individuals is zero if they have identical haplotypes and is one otherwise. For microsatellite data, we chose the chord distance of Cavalli-Sforza and Edwards (1967), which is among the best genetic distances for recovering the correct tree topology according to Takezaki and Nei (1996). Bootstrap values were computed over 2000 replications (Hedges 1992) for all trees, by resampling characters (MP method), loci (ML and distance methods), or individuals within populations (MP, ML, and distance methods). An NJ tree based on the shared allele distance between individual bees was built as described by Estoup et al. (1995a).


mtDNA Data

Among the 544 colonies assayed, 26 different haplotypes were observed: 17 already reported (Garnery et al. 1993, 1995) and nine new haplotypes. Restriction maps (confirmed by sequences, data not shown) of the 26 haplotypes are given in Figure 2. An MP tree of haplotypes is given in Figure 3. This tree corresponds to the 80%-majority-rule consensus tree of more than 9400 equally parsimonious trees. The MP tree confirms the separation of haplotypes in three lineages and the correct assignation of each haplotype to its own lineage.

Haplotype frequencies for each population are given in Figure 4. African (A) haplotypes were found in all African and Iberian samples except in the Basque Country, whereas western European (M) haplotypes were present in all French and Iberian samples except in northern Portugal. A few northern Mediterranean haplotypes (C1) were detected in all French samples and in the Basque Country (two colonies) and Portugal (one colony) samples.

Figure 4 also provides an illustration of intrapopulation variation. Within each population, the mtDNA haplotypes are represented as nodes in a network where adjacent nodes differ by a single DraI site substitution or sequence Q insertion/deletion. When populations contain haplotypes from different lineages, the corresponding networks were not connected because of the numerous differences between haplotypes of these lineages on the remaining part of the mtDNA molecule (Smith 1991; Garnery et al. 1992; see also [ILLUSTRATION FOR FIGURE 3 OMITTED]). Most non-Iberian populations showed networks in which the most frequent haplotype (i.e., the most probable ancestral haplotype in the population; Watterson and Guess 1977; Crandall and Templeton 1993) is related to a few rare satellite haplotypes by one or two mutational events. In African populations, networks are simple, with mainly haplotypes A8 and A9 in northern Morocco and A1 in southern Morocco and Guinea. French and Basque populations also contain closely linked haplotypes, always with high frequencies of haplotypes M4 and M4[prime]. On the contrary, other Iberian populations show complex haplotype networks in both A and M haplotypic classes, which suggests multiple haplotypic origins (Crandall and Templeton 1993; Garnery et al. 1995). In addition, the most frequent haplotypes are either unique (A16 in northern Portugal) or found only in Iberian samples (A2, A11, M3, M7, and M7[prime]), whereas common African haplotypes A1, A8, and A9 are rare in Spain and Portugal.

In agreement with the preceeding observations, the highest haplotype diversities (D) are found in the Iberian samples (Table 1), some of which (Castile and Andalucia) are composed of a mixture of haplotypes from lineages A and M. Even when considering the haplotypes of the two lineages A and M separately, the samples from Castile and northern Portugal still have larger diversities than any African sample for lineage A haplotypes and all the three Spanish samples have higher values of D than any French sample for lineage M haplotypes.

Population trees [ILLUSTRATION FOR FIGURE 5 OMITTED] have been computed with and without the Italian outgroup. No change in the topology occurred when introducing the outgroup. The ML and NJ trees have very similar topologies (there was only a swap of two French populations, Vaucluse and Atlantic Pyrenees). The MP tree differs by the branching of Iberian populations at the tip of the African branch. Because Iberian populations contain a mixture of A and M haplotypes and hence are expected to branch in an intermediate position between A and M populations, the ML and NJ trees were considered as more meaningful than the MP tree. The NJ and MP trees show the existence of five groups: {northern France, Atlantic Pyrenees, Vaucluse, and Basque Country}, {Castile and Andalucia}, {northern Portugal}, {northern Morocco}, and {southern Morocco and Guinea}. Although Iberian populations branch in an intermediate position relative to the North Africa-France axis, the large length of the edges connecting them to the rest of the network prevents them from being considered as simple intermediates between the North African and French populations. Also, the patristic distances among the various Iberian samples are almost as large as the distances between any Iberian and non-Iberian sample.

Microsatellite Data

Allele frequencies for all (locus x sample) combinations are detailed in Table 2. When considering these combinations separately, significant departure from Hardy-Weinberg equilibrium was detected in six instances. However, global tests per population were all negative and these six instances (of 86) concerned five different loci and six different populations, suggesting that significance was most probably reached by chance alone.

Exact tests for linkage disequilibrium resulted in 10 significant values of 371 comparisons (eight microsatellite loci and the COI-COH locus) with three values at the 1% threshold. Although 18 significant values are expected at the 5% level by chance, the distribution of P-values between populations was not homogeneous. The populations from Vaucluse, northern France, and Atlantic Pyrenees included four, two, and one significant values, respectively, with three highly significant values for the population from Vaucluse and one cytonuclear disequilibrium for the population from Atlantic Pyrenees. Nevertheless, multiple probability tests by pair of loci across all populations (Fisher = s method) did not detect any linkage disequilibrium (P [greater than] 0.146). When applying Bonferronni's correction (Rice 1989), the population from Vaucluse contained one significant value between the pairs of loci A24/A88.

African samples displayed higher calculated gene diversities ([H.sub.c]) than European samples (Wilcoxon's signed rank test, P [less than] 0.01 for all 4 X 7 combinations). Among the four African samples, Guinea showed the largest values (Wilcoxon's signed rank test, P [less than] 0.02 for all three comparisons). The average gene diversity (over the eight microsatellite loci) of any African population is more than twice that of European populations (Table 3), as opposed to mtDNA haplotype diversity (Table 1).

In European populations, average calculated gene diversities ([H.sub.c]) were constantly lower than gene diversities ([H.sub.e]) expected under the hypothesis of mutation-drift equilibrium for any mutation model (Table 3). When considering gene diversities at individual loci, this deficiency was significant in every European population (Wilcoxon's signed rank test, P [less than] 0.05 in all 2 models x 7 populations tests). When pooling all 54 polymorphic locus x sample combinations, gene diversity deficiency was found at 45 and 53 loci under the IAM and SMM, respectively. Under the IAM, seven of the nine exceptions concerned locus B124, which is the most polymorphic [TABULAR DATA FOR TABLE 2 OMITTED] locus in European samples and hence is the one that evolves back to equilibrium at the fastest rate.

With any of the three tree reconstruction methods (MP, ML, and NJ on chord distance), there appeared a clear-cut separation of African and European bees, with the connecting branch being supported in 100%, 100%, and 97% of bootstrap replicates, respectively [ILLUSTRATION FOR FIGURE 6 OMITTED]. The addition of the Italian outgroup only brought minor changes in the respective branching of French samples in all three trees and the swapping of the two northern Moroccan samples (northern Rif and southwest Rif) in the MP tree. The dendrogram of individual bees [ILLUSTRATION FOR FIGURE 7 OMITTED] confirms the high differentiation of bees from lineages A and M, which made two pure clusters. The Italian bees (lineage C) make a third cluster, justifying their use as an outgroup. The lower genetic variability of European bees is illustrated by shorter terminal edges in the tree compared to African bees. Within lineages, populations and even subspecies appear mixed except for the Guinean sample, which is rather well clustered. Eleven individual bees (one from southwest Rif, two from Basque Country, two from Atlantic Pyrenees, three from northern France, and three from Vaucluse) branch between Italian bees and the two main A and M clusters [ILLUSTRATION FOR FIGURE 7 OMITTED]. This position in the tree suggests that these bees result from crosses between local A or M populations and imported C bees.

Fisher exact tests for genic and genotypic differentiation also confirm the very high structuring between European and African bees (0.200 [less than] multilocus [] [less than] 0.350 for 4 x 7 pairs of comparisons). Structuring is also significant between all pairs of subspecies (P [less than] 0.008 and P [less than] 0.0002 for genic and genotypic differentiation, respectively). At the population level (within subspecies, i.e., mellifera and iberica), differentiation is significant in some pairs of populations, generally the geographically most distant ones, which suggests an isolation-by-distance mechanism. This hypothesis is strengthened by the close relationship (r = -0.751, P [less than] 0.01) between the number of migrants (estimated from the []-value) and the geographic distance among pairs of European samples [ILLUSTRATION FOR FIGURE 8 OMITTED]. Between pairs of European and African samples, all loci exhibit a high differentiation except locus B124. Within Europe, genetic differentiation is mainly due to loci A28, A113, and B124, whereas loci A88, A28, A113, and A43 structure the populations from Africa.


Our results can be summarized as follows: (1) Along the Guinea-France transect, mitochondrial DNA presents a ratio cline between two highly divergent families of haplotypes (M and A). This cline extends over most of the Iberian Peninsula. Microsatellites indicate a clear-cut disruption between the two continents (Figs. 6, 7). (2) MtDNA haplotype networks [ILLUSTRATION FOR FIGURE 4 OMITTED] are simple in Africa and in France and are much more complex in Spain and Portugal. Microsatellite loci are much more variable in Africa than in western Europe, where data suggest a recent bottleneck that could have affected all studied populations because they appear rather homogeneous (Table 3). (3) French populations are affected by a "genetic pollution" (Figs. 4, 7) due to recent introductions of queens from the C lineage, which could partially disturb phylogenetic and population genetic conclusions. (4) MtDNA and microsatellites are independent markers that are differently affected by sex, have their own pattern of variation and evolutionary rate, and hence are able to reflect historic events occurring at different time periods. These markers are congruent over wide territories. The lack of congruence found in the Iberian Peninsula is unique and requires a specific explanation.

Rejection of the Hypotheses of a Primary or a Secondary Intergradation

The African origin of lineage M, which colonized Europe through the Strait of Gibraltar as suggested by morphology and the MDH ratio cline, might still be compatible with microsatellite data if we consider that a severe bottleneck occurred. Such a bottleneck is not unlikely, as suggested by the inclusion of European alleles within the African spectrum and by the generalized deficiency of genic diversity in all European populations (Tables 2, 3). However, this bottleneck should be recent because the high mutation rate of microsatellites (Weber and Wrong 1993; Gyapay et al. 1994) make them able to recover their variability quickly. However, this hypothesis is incompatible with mtDNA data, regardless of when the bottleneck occurred. The mixture of A and M haplotypes in Spain cannot be a simple modification of an ancestral A + M polymorphism of African populations in the process of colonization because haplotypes M are absent from all African areas studied so far (Hall and Smith 1991; Garnery and Cornuet 1994).

The hypothesis of secondary intergradation was developed to explain the Spanish ratio cline between M and A haplotypes, which are highly divergent (ca. 2.6%; Garnery et al. 1992) and extend on large territories on both sides of the cline. This implies recent contact, because many Spanish A and especially M haplotypes are identical to those found in pure A or M populations. However, this hypothesis raises two difficulties. One is the higher within-lineage haplotype diversity in the Iberian Peninsula, which is unexpected in a contact zone between two sets of marginal populations. The other is with microsatellites, which instead of showing a nuclear mixture (A + M) in the Iberian Peninsula clearly link Spanish and Portuguese populations to other European M populations [ILLUSTRATION FOR FIGURES 6, 7 OMITTED].

An Ancient Divergence and Recent Mitochondrial Introgression

All molecular (mtDNA and microsatellites) data suggest an ancient divergence of A and M lineages over wide territories. However, this does not imply a similar within-group structure.

African populations are dense, have a pronounced migratory behavior, and have lived in a more stable environment. [TABULAR DATA FOR TABLE 3 OMITTED] Effective population sizes are larger in Africa (Estoup et al. 1995a), which explains their higher variability for microsatellite loci (Table 3). However, compared to European populations, African mitochondrial DNA is less variable (Table1). This can be explained by a lower mutation rate in the COI-COII region related to the shorter length of the sequence, as suggested by the significant correlation between the average number of Q sequences and the A haplotype diversity (r = 0.834, P [less than] 0.05). Small sequences offer less targets for site mutations and less possibilities for Q duplication/deletion (Cornuet et al. 1991).

Lineage M haplotype variability is also higher in Iberian than in French populations, where the average length of the COI-COII sequence is about the same and hence the evolutionary rates should not be very different. But Europe experienced many glacial episodes, during which Spain must have been used as a refuge by honeybee populations and most of French populations must have disappeared. It is therefore expected that during recolonization a significant fraction of the variability has been lost. In agreement with this observation, the major French haplotype (M4) is also the most frequent in all Spanish populations where lineage M haplotypes have been observed. Furthermore, the generalized deficit of genic diversity at microsatellite loci (Table 3) also suggests a recent expansion of European populations after a bottleneck.

If larger haplotypes explain higher diversity, it does not explain complex mitochondrial networks for the A haplotypes in Spanish and Portuguese populations. This result suggests multiple origins of colonies from lineage A in Iberian populations, which could be explained by successive African introductions during historic time, such as the Arabian colonization. These recent introductions agree with the fact that no African haplotype has been found in France after a postglacial recolonization.

Mitochondrial DNA and Selection

One question raised by the new hypothesis is how multiple introductions of African lineage colonies have produced significant changes in the mitochondrial but not in the nuclear compartment of the Iberian genome. The lack of foreign influence detected with microsatellites suggest that haplotypes A have spread because they were selectively superior to local haplotypes M and hence escaped the dilution expected with neutral loci (such as microsatellites). The Moroccan situation provides an a contrario confirmation of the putative selective superiority of haplotypes A in circum-Mediterranean countries. When Morocco was under French rule, thousands of mellifera queens were imported into the country (E. H. Mosshine, pers. comm.). Yet, not a single haplotype M could be found in our present-day samples, whereas there is some evidence that microsatellite alleles of M origin still exist there. This evidence includes sequences of alleles at the interrupted microsatellite A113 (PE unpubl. data; for rationale, see Estoup et al. 1995b).

MtDNA is a completely linked molecule. Although the noncoding COI-COII region may be submitted to its own selective forces (e.g., those associated with length) it is quite possible that sequence differences in the important genes of the mitochondrial genome are responsible for significant differences in selective value (Ballard and Kreitman 1995). This differential and allopatric introgression process (Aubert and Solignac 1990) can also be explained by the existence of two mechanisms of gene diffusion specific to social species. One is through mating and is achieved by males; the other is through swarming and is achieved by females. Drones transmit only nuclear genes (such as microsatellites) to unrelated colonies, whereas mtDNA is transmitted only to daughter colonies when swarming. African honeybees present a high propensity for swarming, which can facilitate and accelerate the propagation of their mitochondrial DNA.

Genetic Pollution

The contemporary introductions of queens from the C lineage (especially ligustica queens) in French and Basque populations could affect the results used to establish this new hypothesis. First, the Italian genetic influence in these populations may be responsible for their connection at the bases of the European clade in the MP and ML trees [ILLUSTRATION FOR FIGURE 6 OMITTED]. Second, the introduction of foreign queens may mimic a demographic expansion such as after a bottleneck. However, "genetic pollution" due to imports of ligustica queens is minor overall because it does not distort the close relationship between the number of natural migrants and geographic distance [ILLUSTRATION FOR FIGURE 8 OMITTED] suggesting a simple isolation-by-distance mechanism that affects all Iberian and French populations irrespective of the subspecies and their level of "pollution."

Continuous Phenotypic Variation and Genetic Discontinuity

The existence of an extended clinal variation for morphological characters had prompted the hypothesis of primary intergradation and was congruent with the hypothesis of secondary intergradation. These morphological clines extend from the equator to the Polar Circle, that is, over distances that are much larger than the formerly postulated transition/contact zone. These clines exist in several places over the species' entire distribution, including the New World where honeybees were introduced only recently. They are considered as an adaptation to climatic and ecological constraints (Ruttner 1988; Sheppard et al. 1991; Oldroyd et al. 1994). If this is the case, the Iberian example shows that apparently continuous clines are not affected by a major geographical and genetic discontinuity at the Strait of Gibraltar and that selective forces can shape continuous phenotypic variation from highly diverged genetic backgrounds.


This work shows that the analysis of various characters, even when they vary in parallel, may still lead to erroneous conclusions. It also illustrates the fact that the pertinence of characters is more important than their number. In evolutionary studies, selective neutrality should be the primary quality of genetic markers for the historic and demographic reconstruction of populations. Although microsatellites have their own limits (see Estoup et al. [1995b] for homoplasy and Rubinsztein et al. [1995] for evolution directionality), their high level of variability and their putative selective neutrality make them useful tools for assessing the evolutionary history of populations. In the present case, microsatellites provided new insights on the complex history of several honeybee subspecies.


We are indebted to M. Harry, G. Fert, E. H. Mosshine, A. Daoudi, and R. Borneck for honeybee samples. We wish to thank A. Estoup for helpful suggestions. This work was partly supported by a grant (no. 96006) from the Bureau des Ressources Genetiques.


ALPATOV, W. W. 1929. Biometrical studies on variation and races of honeybee. Q. Rev. Biol. 4:1-58.

AUBERT, J., AND M. SOLIGNAC. 1990. Experimental evidence for mitochondrial DNA intogression between Drosophila species. Evolution 44:1272-1282.

BALLARD, J. W. O., AND M. KREITMAN. 1995. Is mitochondrial DNA a strictly neutral marker? Trends Ecol. Evol. 10:485-488.

CAVALLI-SEORZA, L. L., AND A. W. F. EDWARDS. 1967. Phylogenetic analysis: Models and estimation procedures. Am. J. Hum. Genet. 19:233-257.

CHAKRABORTY, R., AND L. JIN. 1993. A unified approach to study hypervariable polymorphisms: statistical considerations of determining relatedness and population distance. Pp. 153-175 in S. D. J. Pena, R. Chakraborty, J. T. Epplen, and A. J. Jeffreys, eds. DNA fingerprinting: state of the science. Birkhauser Verlag Basel, Switzerland.

CORNUET, J.-M. 1982. The MDH polymorphism in some west mediterranean honeybee populations. IUSSI congress, Boulder, CO.

-----. 1983. Reproduction, genetique et selection de l'abeille. Bull. Tech. Apic. 10:13-36.

CORNUET, J.-M., AND J. FRESNAYE. 1989. Etude biometrique de colonies d'abeilles d'Espagne et du Portugal. Apidologie 20:91-101.

CORNUET, J.-M., AND L. GARNERY. 1991. Mitochondrial DNA variability in honeybees and its phylogeographic implications. Apidologie 22:627-642.

CORNUET, J.-M., AND G. LUIKART. 1996. Description and power analysis of two tests for inferring recent population bottlenecks from allele frequency data. Genetics 144:2001-2014.

CORNUET, J.-M., J. FRESNAYE, AND L. TASSENCOURT. 1975. Discrimination et classification de populations d'abeilles a partir de caracteres biometriques. Apidologie 6:145-187.

CORNUET, J.-M., J. FRESNAYE, AND P. LAVIE. 1978. Etude biometrique de deux populations d'abeilles cevenoles. Apidologie 9: 41-55.

CORNUET, J.-M., J. ALBISETTI, N. MALLET, AND J. FRESNAYE. 1982. Etude biometrique d'une population d'abeilles landaises. Apidologie 13:3-13.

CORNUET, J.-M., A. DOUADI, H. MOSSHINE, AND J. FRESNAYE. 1988. Etude biometrique de populations d'abeilles marocaines. Apidologie 19:355-366.

CORNUET, J.-M., L. GARNERY, AND M. SOLIGNAC. 1991. Putative origin and function of the COI-COII intergenic region of Apis mellifera L. mitochondrial DNA. Genetics 1128:393-403.

CORNUET, J.-M., B. P. OLDROYD, AND R. H. CROZIER. 1995. Unequal thermostability of allelic forms of malate dehydrogenase in honeybees. J. Apic. Res. 34:45-47.

CRANDALL, K. A., AND A. R. TEMPLETON. 1993. Empirical tests of some predictions from coalescent theory with applications to intraspecific phylogeny reconstruction. Genetics 134:959-969.

ESTOUP, A., L. GARNERY, M. SOLIGNAC, AND J.-M. CORNUET. 1995a. Microsatellite variation in honey bee (Apis mellifera L.) populations: hierarchical genetic structure and test of the infinite allele and stepwise mutation models. Genetics 140:679-695.

ESTOUP, A., C. TAILLIEZ, J.-M. CORNUET, AND M. SOLIGNAC. 1995b. Size homoplasy and mutational processes of interrupted micro-satellites in apidae species, Apis mellifera and Bombus terrestris. Mol. Biol. Evol. 12:1074-1084.

FELSENSTEIN, J. 1993. PHYLIP: phylogeny inference package. Vers. 3.5. University of Washington, Seattle, WA.

GARNERY, L., AND J.-M. CORNUET. 1994. Inter and intra lineage mitochondrial DNA variation in the honeybee Apis mellifera. 12th IUSSI Congress, Paris.

GARNERY, L., D. VAUTRIN, J.-M. CORNUET, M. SOLIGNAC. 1991. Phylogenic relationships in the genus Apis inferred from mitochondrial DNA sequence data. Apidologie 22:87-92.

GARNERY, L., J.-M. CORNUET, AND M. SOLIGNAC. 1992. Evolutionary history of the honey bee Apis mellifera inferred from mitochondrial DNA analysis. Mol. Ecol. 1:145-154.

GARNERY, L., M. SOLIGNAC, G. CELEBRANO, AND J.-M. CORNUET. 1993. A simple test using restricted PCR-amplified mitochondrial DNA to study the genetic structure of Apis mellifera. Experientia 49:1016-1021.

GARNERY, L., H. MOSSHINE, B. P. OLDROYD, AND J.-M. CORNUET. 1995. Mitochondrial DNA variation in Moroccan and Spanish honey bee populations. Mol. Ecol. 4:465-471.

GYAPAI, G., J. MORISSETTE, A. VIGNAL, C. DIB, C. FIZAMES, P. MILLASSEAU, S. MARC, O. BERNARDI, M. LATHROP, AND J. WEISSENBACH. 1994. The 1993-94 Genethon human genetic linkage map. Nat. Genet. 7:246-339.

HALL, H. G., AND D. R. SMITH. 1991. Distinguishing African and European honeybee matrilines using amplified mitochondrial DNA. Proc. Nat. Acad. Sci. USA 88:4548-4552.

HEDGES, S. B. 1992. The number of replications needed for accurate estimation of the bootstrap P value in phylogenetic studies. Mol. Biol. Evol. 9:366-369.

HEPBURN, H. R., AND S. E. RADLOFF. 1996. Morphometric and pheromonal analyses of Apis mellifera L. along a transect from Sahara to the Pyrenees. Apidologie 27:35-45.

KOCHER, T. D., W. K. THOMAS, A. MEYER, S. V. EDWARDS, F. X. VILLABLANCA, A. C. WILSON. 1989. Dynamics of mitochondrial DNA evolution in animals: amplification and sequencing with conserved primers. Proc. Nat. Acad. Sci. USA 86:6196-6200.

NEI, M. 1978. Estimation of average heterozygoty and genetic distance from a small number of individuals. Genetics 89:145-163.

NEI, M., AND F. TAJIMA. 1981. DNA polymorphism detectable by restriction endonucleases. Genetics 97:583-590.

OLDROYD, B. P., D. ROWE, J.-M. CORNUET, T. E. RINDERER, AND R. H. CROZIER. 1994. Racial admixture of Apis mellifera in Tasmania, Australia: similarities and differences with natural hybrid zones in Europe. Heredity 74:315-325.

ORANTES-BERMEJO, F. J., AND P. GARCIA-FERNANDEZ. 1995. Morphological variability of Apis mellifera iberica in different apiaries of southern Spain. J. Apic. Res. 34:13-30.

RAYMOND, M., AND F. ROUSSET. 1995. Genepop (version 1.2): population genetics software for exact test and ecumenism. J. Hered. 86:248-250.

RUBINSZTEIN, D. C., W. AMOS, J. LEGGO, S. GOODBURN, S. JAIN, S. H. LI, R. L. MARGOLIS, C. A. Ross, AND M. A. FERGUSSON-SMITH. 1995. Microsatellite evolution: evidence for directionality and variation in rate between species. Nat. Genet. 10:337-343.

RUTTNER, F. 1988. Biogeography and taxonomy of honeybees. Springer-Verlag, Berlin.

RUTTNER, F., L. TASSENCOURT, AND J. LOUVEAUX. 1978. Biometrical-statistical analysis of the geographic variability of Apis mellifera L. Apidologie 9:363-381.

SAITOU, N., AND M. NEI. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425.

SHEPPARD, W. S., T. E. RINDERER, J. A. MAZZOLI, J. A. STELZER, AND H. SHIMANUKI. 1991. Gene flow between African and European-derived honeybee populations in Argentina. Nature. 349: 782-784.

SLATKIN, M. 1993. Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47:264-279.

SMITH, D. R. 1991. African bees in Americas: insight from biogeography and genetic. Trends Ecol. Evol. 6:17-21.

SMITH, D. R., AND T. C. GLENN. 1995. Allozyme polymorphisms in Spanish honeybees (Apis mellifera iberica). J. Hered. 86:12-16.

SMITH, D. R., M. F. PALOPOLI, B. R. TAYLOR, L. GARNERY, J.-M. CORNUET, M. SOLIGNAC, AND W. M. BROWN. 1991. Geographical overlap of two mitochondrial genomes in spanish honeybees (Apis mellifera iberica). J. Hered. 82:96-100.

SNEDECOR, G. W., AND W. G. COCHRAN. 1978. Statistical methods. Iowa State Univ. Press, Ames, IA.

SWOFFORD, D. L. 1989. PAUP: phylogenetic analysis using parsimony. Rel. 3. Illinois Natural History Survey, Champaign, IL.

TAKEZAKI, N., AND M. NEI. 1996. Genetic distances and reconstruction of phylogenetic trees from microsatellite DNA. Genetics 144:389-399.

WATTERSON, G. A., AND H. A. GUESS. 1977. Is the most frequent allele the oldest. Theor. Popul. Biol. 11:141-160.

WEBER, J. L., AND C. WRONG. 1993. Mutation of human short tandem repeats. Hum. Mol. Genet. 2:1123-1128.
COPYRIGHT 1998 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Franck, Pierre; Garnery, Lionel; Solignac, Michel; Cornuet, Jean-Marie
Date:Aug 1, 1998
Previous Article:The genetic basis of the trade-off between calling and wing morph in males of the cricket Gryllus firmus.
Next Article:Test for ancient species flocks based on molecular phylogenetic appraisals of Sebastes rockfishes and other marine fishes.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters