The initiation of foraging during the life course of honeybee workers is of central interest to understanding the division of labor in social insects, a central theme in sociobiology and behavioral research. It also provides one of the most complex phenotypic traits in biological systems because of the interaction of various external, social, and individual factors. This study reports on a comprehensive investigation of the genetic architecture of the age of foraging initiation in honeybees. It comprises an estimation of genetic variation, the study of candidate loci, and two complementary quantitative trait loci (QTL) maps using two selected, continually bred lines of honeybees. We conclude that considerable genetic variation exists between the selected lines for this central life history component. The study reveals direct pleiotropic and epistatic effects of candidate loci (including previously identified QTL for foraging behavior). Furthermore, two maps of the honeybee genome were constructed from over 400 AFLP markers. Both maps confirm the extraordinary recombinational size of the honeybee genome. On the basis of these maps, we report four new significant QTL and two more suggestive QTL that influence the initiation of foraging.

---

RECENT developments in quantitative genetics and genomics allow for genetic analyses of increasingly complex traits in a variety of organisms. This development is of unprecedented value for all biological disciplines, as most phenotypic variation of interest is determined by genetic and environmental components that often interact in elaborate ways. In addition to a quantification of the overall genetic variation, single gene effects and their interactions need to be identified because the sum of the constituent factors might be not only greater but also different from its parts (BRODIE 2000). Behavioral and life-history traits can be regarded as prime examples of complex, polygenic traits that often cannot be reduced to simple components (FINCH and ROSE 1995).

Social insects constitute a particularly interesting system to consider the causation of such polygenic traits because their social environment adds an additional level of complexity that is itself partially under genetic control (WOLF 2000; PANKIW et al. 2002). The honeybee (Apis mellifera L.) has become a model in several research areas for a variety of reasons (ROBINSON et al. 1997; MENZEL 1999; PAGE et al. 2002). One of the most important aspects of honeybee biology is the intriguing and well-described temporal division of labor among its workers: Young workers specialize in particular duties in the nest, while older workers forage for resources outside the nest (WINSTON 1987).

This transition of individual honeybees from hive bees to foraging initiation has been the focus of much theoretical (BESHERS and FEWELL 2001) and empirical (e.g., ROBINSON 2002) research effort. While successful modeling can fruitfully guide further research, model interpretation ultimately requires knowledge of the underlying causation of this behavioral phenomenon (ROBINSON et al. 1997). Some progress in understanding the mechanisms of foraging initiation has been made by investigating hormones (SULLIVAN et al. 2000), neurotransmitters (SCHULZ and ROBINSON 2001), and candidate genes (reviewed by ROBINSON 2002). However, an extensive genetic dissection of this central life-history trait has never been attempted in spite of its high potential to reveal genetic structure and causative candidates.

The onset of foraging is a complex trait with interacting endogenous developmental processes (ROBINSON et al. 1994), potential neurogenic (TOMA et al. 2000) or neuro-endocrine pacemakers (SULLIVAN et al. 2000; SCHULZ and ROBINSON 2001), and social (pheromonal) influences (HUANG and ROBINSON 1992; PANKIW et al. 1998; LECONTE et al. 2001). Furthermore, environmental conditions that influence the colony status, such as food shortage (SCHULZ et al. 1998), influence the individual onset of foraging. It has been shown that individual factors, such as octopamine, interact with other factors in a nontrivial manner (BARRON et al. 2002; SCHULZ et al. 2002). Such interactions may occur even between different levels of organization (cell-organ/individual-colony; PAGE and ERBER 2002) and constitute the backbone of theoretical models for the onset of foraging (HUANG and ROBINSON 1992; BESHERS and FEWELL 2001; AMDAM and OMHOLT 2003). Thus, an array of parameters makes many genes potentially influential. Genetic variability for the age at foraging initiation has been shown repeatedly in honeybees (CALDERONE and PAGE 1988; ROBINSON et al. 1989; PAGE et al. 1991; GIRAY et al. 2000; BRILLET et al. 2002).

We studied the phenomenon of foraging initiation using a comprehensive, quantitative genetic approach, exploiting pronounced differences in the age of foraging initiation (or age of first foraging: AFF) between two lines of honeybees. These lines had previously been found to differ significantly with respect to AFF (PANKIW and PAGE 2001), although they were selected on the basis of their pollen-hoarding behavior (PAGE and FONDRK 1995). Bees selected for high pollen-hoarding behavior were found to initiate foraging earlier than both unselected bees and bees selected for low pollen-hoarding behavior (PANKIW and PAGE 2001).

This correlated selection response, which pertains to other traits such as sucrose sensitivity (PAGE et al. 1998) and learning behavior (SCHEINER et al. 2001), prompts the question whether the behavioral and life-history differences between the selected lines are genetically correlated (by close linkage or pleiotropy) or the two selected lines have been fixed by chance for alternative alleles at different loci (each independently influencing a single trait).

In this study, we generated four experimental groups with different genetic composition from the two selected lines to investigate the genetic architecture of the AFF in worker honeybees. Our specific aims were: (1) to quantify overall genotypic effects, (2) to test for genetic effects of genetic markers linked to previously identified QTL affecting foraging behavior (HUNT et al. 1995; PAGE et al. 2000) and to the candidate genes Amper (period ortholog, GenBank accession no. AF159569) and Amfor ("PKG" ortholog, GenBank accession no. AF469010) that have both been associated with the initiation of foraging (TOMA et al. 2000; BEN-SHAHAR et al. 2002; ROBINSON 2002), and (3) to construct genomic linkage maps in reciprocal backcross populations to screen the honeybee genome systematically for new QTL that affect the age of onset of foraging behavior. To understand the genetic architecture of complex traits, the study of such multiple segregating populations is far more informative than results from a single experimental population (ASíNS 2002).

Experimental groups:

A high and a low pollen-hoarding line set up by PAGE and FONDRK (1995) were used as genetic sources for the experiment. These lines were derived from strains established in 1990 and maintained under continuous, bidirectional selection for pollen-hoarding behavior. At the beginning of the experiment, the strains were in their eighteenth generation, maintained by circular inbreeding among lines and three outcrossings to commercial hives of similar (high pollen-hoarding or low pollen-hoarding) phenotype.

We generated hybrid supersister queens (PAGE and LAIDLAW 1988) by instrumentally inseminating a virgin queen from a low line ("C") with semen of a single drone from one high line ("Q1"). Eggs of this queen were grafted into queen cups to produce hybrid queen offspring (LAIDLAW and PAGE 1997). Subsequently, these hybrid queens were instrumentally inseminated with sperm of either a high-line male or a low-line male. Two of these queens were selected to generate the high backcross (HBC) and low backcross populations (LBC). Worker offspring of the corresponding high- and low-line queens (eighteenth generation) represented our two other experimental groups (high and low, Figure 1).

View larger version (15K): In this window In a new window Download PPT slide   FIGURE 1.— Breeding scheme used in the experiment to generate two reciprocal backcross populations and two populations from the original source lines (PAGE and FONDRK 1995). HBC, high backcross; LBC, low backcross; High, high pollen hoarding; Low, low pollen hoarding.

 

Experiment:

All four source queens were simultaneously caged on empty comb within their respective colonies to induce maximal, simultaneous egg-laying behavior. Twenty-four hours prior to emergence, these combs were transferred into individual cages in temperature- (34°) and moisture (50% relative humidity)-controlled incubators. Newly emerged bees of each source were color marked on the thorax with a unique enamel color and transferred within 6 hr to a common, unrelated bee colony (common hive environment).

Returning foragers from this host colony were monitored daily for 2 hr during peak foraging (PANKIW and PAGE 2001). Each marked returning forager was captured and assigned a unique number, and her date of capture was recorded (which corresponded to her AFF). Our observational period proved long enough to capture the vast majority of marked bees flying on any specific day, and we avoided collection during orientation flights (PANKIW and PAGE 2001).

Evaluations of candidate markers:

DNA was purified from whole bees with a CTAB lysis and a single phenol/chloroform extraction (HUNT and PAGE 1995). After quantification of the resuspended DNA with a DNA fluorometer (Hoefer Scientific Instruments, San Francisco), the DNA concentration was adjusted to 100 ng/µl. Specific primers were used to PCR amplify previously identified sequence-tagged-site (sts) markers (HUNT and PAGE 1995; PAGE et al. 2000) that are linked to identified QTL (pln1, pln2, and pln3) or to candidate genes (Amfor and Amper). The QTL markers were within 15 cM of their respective "QTL-peak," and the candidate gene markers were designed from the immediate sequence of the respective gene.

Table 1 lists these candidate loci with information on primers and PCR conditions. When PCR amplification did not result in a presence/absence polymorphism, the amplified DNA sequences were screened for polymorphisms (segregating markers) with >20 restriction endonucleases. The restriction enzymes used in the final analyses are also given in Table 1. All candidate loci were evaluated on 20 x 25-cm agarose gels [0.64% agarose, 1% synergel (Diversified Biotech), 0.5x TBE] and ethidium bromide staining. We tested each marker's effect on the AFF by a two-way ANOVA (marker genotype x backcross population) and corresponding posthoc tests. Furthermore, we tested for epistasis among candidate loci by complete multiway ANOVAs.

View this table: In this window In a new window   TABLE 1 The five candidate genome regions investigated for their effect on the age at foraging initiation in both backcross populations

 

QTL analyses:

AFLP marker generation:

A mapping population of 182 individuals from the HBC group and a second mapping population of 94 individuals from the LBC group were generated, choosing individuals with extreme phenotypes in both cases. The AFLP core reagent kit (Invitrogen Life Technologies) was used according to the manufacturer's recommendations (with reaction volumes reduced to 50%) for the double restriction (EcoRI and MseI) and ligation of the adaptors. The preamplification involved 24 cycles of denaturation (94°/60 sec), annealing (56°/60 sec), and elongation (72°/60 sec). The product was diluted 1:50 to serve as template for a second, selective PCR. Selective primers contained two specific nucleotides and their sequences are given in the APPENDIX. The EcoRI primers were labeled with [-³³P]ATP and used at a final concentration of 85 ng/ml. The unmodified MseI primers were used at 625 ng/ml.

The products of the selective amplifications were run out on denaturing, 6% polyacrylamide gels (35 cm x 45 cm x 0.4 mm), along with radioactively labeled Sequamark size marker (Research Genetics, Hunstville, AL). Gels were dried on filter paper and exposed for 24–96 hr to Biomax MR film (Kodak). Polymorphic bands were identified visually, sized, and scored for all individuals. Finally, all films were rescored once to check for scoring errors. For most markers, linkage phase was established by comparison with the genotype of the grandfather of the mapping populations (Figure 1).

Map construction:

Linkage groups in both marker data sets were identified with Mapmaker 3.0b (LANDER and BOTSTEIN 1989). We used the Kosambi mapping function and 37.5 cM/LOD 3 as linkage criteria between markers (HUNT and PAGE 1995). Linkage groups identified by two-point analysis were subsequently scrutinized by multipoint analysis and exhaustive tests of local map permutations to find the best marker order within linkage groups (LINCOLN et al. 1993). An ordered data set was generated and the data checked for potential PCR artifacts indicated by double crossovers (HUNT and PAGE 1995).

The linkage maps from the HBC and LBC were compared to establish homology of markers and linkage groups. The nature of AFLP markers (codominant only if a length polymorphism is involved, dominant in the more common case of presence/absence polymorphism) makes this process difficult and we resorted to a third map, constructed from the haploid male offspring of one of the hybrid queens (our unpublished results). In theory, both the HBC and the corresponding LBC marker can be scored in the haploid drones to establish homology between corresponding markers. Thus, we used the linkage groups of the male map to determine homology for HBC and LBC linkage groups without match in the direct LBC-HBC map comparison.

We used two criteria to establish homology between linkage groups in the HBC and LBC maps: (1) Two linkage groups were regarded as homologous when both contained three or more markers in the same order that were generated with, respectively, the same primer combination and (2) two linkage groups were regarded as homologous when both contained two or more markers that had exact matches in the same linkage group in the male map.

QTL analysis:

All markers were initially screened for an association with the age of first foraging using simple one-way ANOVAs. As data were bimodal and significance levels would have to be corrected for multiple comparisons, these tests were used only to sort markers according to their effect size. Additionally, we ordered the HBC and LBC marker sets according to the results of the linkage mapping and analyzed both data sets with the computer programs MapQTL 4.0 (VAN OOIJEN et al. 2002) and QTL Cartographer (BASTEN et al. 1994, 2002). Interval mapping (JANSEN 1993) was used, and as both programs resulted in almost identical genome profiles of the LOD score, only the results of MapQTL are shown here. The genome-wide significance threshold for LOD scores was adopted from HUNT et al. (1995)(see also LANDER and BOTSTEIN 1989), and these scores were compared to empirically determined values (CHURCHILL and DOERGE 1994).

Overall genetic differentiation:

The age of first foraging was significantly different among the high (mean 16.1 ± SD: 5.9; n = 185), HBC (17.1 ± 7.8; n = 623), LBC (20.8 ± 6.1; n = 511), and low (26.7 ± 5.5; n = 165) groups (F_(3.1480) = 109.08, P < 0.0001, Figure 2). Tukey's HSD posthoc test revealed highly significant differences (P < 0.001) among all individual groups except between the high and the HBC groups (NS). The proportion of phenotypic variation attributable to genotype was measured over all groups by regression analysis (r² = 0.16, F_(1.1482) = 277.55, P < 0.0001; Figure 2). When only the two original lines (high and low genotype groups) were included, r² = 0.46 (F_(1.348) = 300.19, P < 0.0001). The regression equations were Y [days] = –0.40X [days] + 24.1 [days] and Y [days] = –0.68X [days] + 26.7 [days], respectively. Y represents the AFF and X the proportion of genome that originates from the high line.

View larger version (33K): In this window In a new window Download PPT slide   FIGURE 2.— The age at the onset of foraging (age of first foraging) strongly differed between the experimental groups. The cumulative hazard plot of foraging initiation is shown. Inset: Means and SD of the AFF (in days) of the four groups and the regression of AFF on genotype.

 
Although both backcrosses matured faster than expected on the basis of a linear additive model (_HBC = 1.77 days, _LBC = 3.41 days), this deviation was not significant. As expected, the random recombination of genomes in both backcross populations resulted in higher phenotypic variance than in the source strains (high = 35.08, low = 30.21, HBC = 61.22, LBC = 36.69, Figure 2).

Candidate genes:

Effects of single candidates in both backcross populations were evaluated by two-way ANOVA (marker genotype x backcross population). The effect of backcross population was highly significant (F-values > 13, P-values < 0.001) in all statistical tests, with the HBC group having a lower AFF (17.10 ± 6.06) than the LBC group (20.76 ± 7.82). For pln1, a significant interaction between marker genotype and backcross population was found (F_(1.725) = 7.46, P = 0.0065). This was due to an effect of the pln1 marker in the HBC (F_(1.389) = 7.22, P = 0.0075), but not in the LBC group (F_(1.344) = 0.95, P = 0.3293). No other marker effects were detected (pln2 overall, F_(1.707) = 0.73, P = 0.394; interaction, F_(1.707) = 0.57, P = 0.452; pln3 overall, F_(1.657) = 3.454, P = 0.0635; interaction, F_(1.657) = 0.136, P = 0.713).

Similarly, the Amfor marker results did not have a significant effect (overall, F_(1.696) = 3.272, P = 0.0709; interaction, F_(1.696) = 0.303, P = 0.582). The Amper marker did not have any effect in the LBC group (F_(1.293) = 0.138, P = 0.711). It could not be evaluated in the HBC group due to lack of variability.

The multifactorial ANOVA did not indicate any epistatic effects in the HBC, but revealed interaction effects in the LBC population: A significant two-way interaction was found between the pln1 and the pln3 marker (F_(1.165) = 4.966, P = 0.0272), together with two significant three-way interactions, pln1 x pln2 x pln3 (F_(1.165) = 4.476, P = 0.0359) and pln1 x pln2 x Amper (F_(1.165) = 8.382, P = 0.0043).

Genomic maps:

Overall, 438 AFLP markers were generated in the HBC. Of these, 387 markers could be scored reliably and were linked to one of 35 separate linkage groups (three or more linked markers). The total map size was 3897 cM. In the "low" backcross, 435 markers were generated. These were assembled into 38 linkage groups that comprised 396 markers. The total map size was 3702 cM and 39 markers were unlinked. When adding 37.5 cM for every linkage group in excess of the expected 16 (SANDERSON and HALL 1948), the map size estimates were increased to 4610 and 4527 cM, respectively. Comparisons between the HBC, LBC, and the male map allowed us to combine several linkage groups in one map that had homologies to a single linkage group in one of the other maps. Thus, 16 linkage groups in the HBC map were condensed to 7, and 11 groups in the LBC were combined into 5, resulting in 26 and 32 linkage groups, respectively.

The map comparisons also revealed a reasonable degree of homology between the HBC and LBC maps. Fifteen linkage groups (containing 320 markers in the HBC and 299 markers in the LBC) could be established as homologous between the two backcross maps. This corresponded to 79 and 75% of the total genetic map of the HBC and LBC, respectively.

QTL analysis:

The preliminary ANOVA analyses in the HBC and LBC indicated a few markers of strong effect on the AFF of the focal honeybee workers. Table 2 lists, for both backcrosses, the three markers with the strongest effect on AFF. Three of these six most promising markers did not map to any linkage groups and are therefore not reflected in the QTL described below.

View this table: In this window In a new window   TABLE 2 The three markers with highest F-scores (correlated best with the age of first foraging) in the ANOVA analyses in the HBC and LBC

 
Interval mapping revealed two major QTL for age of first foraging in the HBC (AFF1 and AFF2, Figure 3) and two in the LBC (AFF3 and AFF4, Figure 4). The LOD score of AFF1 was 4.2 and that of AFF2 was 3.5, explaining 16.3 and 14.2% of the total variance, respectively. AFF3 had a LOD score of 3.3 and AFF4 of 3.2, explaining 23.1 and 17.0% of the total variance, respectively. On the basis of the calculated significance threshold of 3.0 (LANDER and BOTSTEIN 1989; HUNT et al. 1995), these four QTL are significant. However, bootstrapping indicated a LOD threshold for significance (3.2 for HBC and 3.4 for LBC) slightly higher than the theoretically calculated one.

View larger version (21K): In this window In a new window Download PPT slide   FIGURE 3.— Two regions in the high backcross showed LOD scores higher than the significance threshold. The x-axis represents the linkage group with recombinational distances between markers. We called these regions (A) AFF1 and (B) AFF2 to indicate their effect on the age of first foraging.

 

View larger version (26K): In this window In a new window Download PPT slide   FIGURE 4.— Two regions in the low backcross showed significant LOD scores. These were called (A) AFF3 and (B) AFF4, and they have different genomic locations from the two QTL shown in Figure 3.

 
Both QTL in the HBC had effects in the expected direction; i.e., the allele for fast behavioral maturation was inherited from the high line. For the two QTL identified in the LBC, the opposite was true: In both cases the allele for fast behavioral maturation was inherited from the low line (Table 3). The two QTL identified in the HBC were located on different linkage groups from the two significant QTL in the LBC. Therefore, the four reported QTL are independent and identify separate genomic regions. The maximum LOD score of the homologous region in the opposite backcross was 0.00 for AFF1, 0.27 for AFF3, and 0.17 for AFF4. For AFF2, no homologous region was identified in the LBC map.

View this table: In this window In a new window   TABLE 3 Effects (mean ± SD) of the ALFP markers most closely linked to the LOD peak of the four significant QTL identified in this study

 
Additionally, we found one putative QTL in the HBC, characterized by marker E5M3323 (LOD score 2.5), and one in the LBC, associated with the marker E9M1363 (LOD score 2.0). Both loci are unrelated to any of the reported significant QTL because they are located on different linkage groups (Figure 5).

View larger version (24K): In this window In a new window Download PPT slide   FIGURE 5.— One additional putative QTL was detected in each backcross. The LOD scores of these QTL did not reach the significance threshold but both have a notable effect on AFF, explaining 9.8 and 13.3% of the total variation in HBC and LBC, respectively.

 

Many aspects of the division of labor, central to the organization of social insect colonies, can be explained by individual response thresholds (BESHERS and FEWELL 2001) that are linked to genotypic differences (e.g., ROBINSON et al. 1989). The onset of foraging is probably the most important and most prominent event in the age-related division of labor in honeybees. PANKIW and PAGE (2001) reported significant genotypic effects on the initiation of foraging between two lines of honeybees from A. mellifera, which were selected on the basis of pollen-hoarding behavior (PAGE and FONDRK 1995). This study confirmed and significantly extended their results by quantifying the genetic differentiation of these lines and studying the underlying genetic architecture. Our results provide the foundation for the future characterization of genes with major causal effects on the rate of behavioral maturation in honeybees.

We quantified the proportion of the variance that is attributed to differences in genotype and evaluated dominance effects by calculating the dominance deviation (FALCONER and MACKAY 1996) in both reciprocal backcrosses. The quantification of genetic and environmental effects in a common hive environment is problematic because social interactions may cause genotype x genotype interactions (CALDERONE and PAGE 1991; WOLF 2000; RüPPELL et al. 2001; PANKIW et al. 2002). Although social interactions influence the rate of behavioral maturation (HUANG and ROBINSON 1992, 1996), we believe our estimate is little affected by G x G interaction because the experimental groups constituted only a very small part of the social environment (i.e., altogether 1500 bees were introduced into a hive of 30,000 bees).

The amount of genetic variation in AFF is fairly high for a central life-history parameter (MOUSSEAU and ROFF 1987). This could be due to the fact that a certain amount of variability in the AFF is advantageous at the colony level and thus genetic variation is maintained (PAGE et al. 1995). On the other hand, the pronounced genotypic differences in behavioral ontogeny are a correlated response to selection on the amount of pollen hoarding (PAGE and FONDRK 1995; PANKIW and PAGE 2001) and adaptive explanations for the maintenance of genetic variability do not need to be invoked. The genetic correlation between AFF and foraging behavior (see also PANKIW 2003) is not trivial. It suggests that the artificial selection regime has modified fundamental pathways of general influence on behavior that are constrained by pleiotropy and epistasis (see below). Incorporating the backcrossed populations expectedly decreased the amount of phenotypic variation that can be explained by genotype because of the inherent genetic heterogeneity of the backcrosses. As neither the HBC nor the LBC population demonstrates a bimodal distribution of the age at onset of foraging, we could further conclude that multiple segregating factors were involved in the pronounced genetic differentiation of the AFF between the high and low line.

Several molecular pathways have been suggested to control, or at least influence, the rate of behavioral development in honeybee workers and thus their age at foraging initiation (HUANG and ROBINSON 1992, 1996; SCHULZ and ROBINSON 1999, 2001; SULLIVAN et al. 2000; TOMA et al. 2000; BEN-SHAHAR et al. 2002; BLOCH et al. 2002; AMDAM and OMHOLT 2003; HUMPHRIES et al. 2003). In this study, we failed to find a significant direct effect of two genetic markers located near the candidate genes Amfor (BEN-SHAHAR et al. 2002) and Amper (TOMA et al. 2000) that have been suggested to be associated with or involved in the initiation of foraging (ROBINSON 2002).

Our results do not disprove the involvement of Amper or Amfor in the initiation of foraging. However, the pronounced differences in the rate of behavioral maturation between the two investigated strains cannot be explained by variability in the genomic region of Amper or Amfor but must be predominantly due to genetic variation in other genomic regions. The gene Amfor has been identified by sequence similarity to the foraging gene (For) in Drosophila melanogaster and it encodes a cGMP-dependent protein kinase that has potential for a variety of cellular functions (BEN-SHAHAR et al. 2002). It has been implicated in the initiation of foraging mainly because its expression increases independently of age with the onset of foraging (BEN-SHAHAR et al. 2002). On the basis of our results we cannot support the hypothesis that Amfor influences the initiation of foraging, but this may be due to a lack of effectual genetic variability in our experimental crosses.

However, the Amper marker contributes to the variability in the onset of foraging in a complex way, interacting with pln1 and pln2. It has been shown previously that the onset of foraging is normally preceded by an upregulation of the activity of the Amper gene (TOMA et al. 2000). The increase in gene activity also correlates with the rise of endogenous circadian rhythmicity (TOMA et al. 2000). The interaction effect demonstrated in this study fits well with the view that Amper could facilitate the effect of other factors that lead to the initiation of foraging. For example, rhythmic bees that are more active during the day are more exposed to recruitment from incoming foragers. Under this scenario, it seems plausible that Amper has a modulatory effect on the age of first foraging that can be rendered insignificant by extreme social conditions (BLOCH et al. 2001).

We regarded the QTL pln1, pln2, and pln3 that were identified previously for their effect on pollen hoarding and foraging (HUNT et al. 1995; PAGE et al. 2000) as potential candidates to influence the rate of behavioral maturation. Our rationale for this hypothesis was the phenotypic correlation of pollen hoarding and fast maturation (PANKIW and PAGE 2001; PANKIW 2003). One of these loci, pln1, showed a direct pleiotropic effect on the age of first foraging. The allele inherited from the "low" line caused individuals to forage, on average, 2.7 days earlier. The fact that the "fast maturation" allele is inherited from the "slow maturation" line may seem counterintuitive, but such results are fairly common in artificial selection programs (e.g., BURKE et al. 2002). Correspondingly, the pln1 allele from the low line exerts an effect that is contrary to the phenotype of its strain of origin on the amount of stored pollen, pollen-foraging behavior (HUNT et al. 1995; PAGE et al. 2000), and sensitivity to sucrose (our unpublished data). This result suggests a genetic architecture with nonadditive gene interactions (OMHOLT et al. 2000; see below) and additional factors with opposite effects (e.g., AFF1 and AFF2). A complex genetic architecture with nonlinear interactions is further supported by several (epistatic) interaction effects among our candidate loci found in the LBC. Epistasis may be the rule in natural genomic architectures of complex traits (TEMPLETON 2000). However, its detection is severely limited in most genome-wide QTL studies (FRANKEL and SCHORK 1996). This is also true for our QTL maps in the HBC and LBC, in which we did not test for any epistasis between the newly generated markers. Thus, we want to suggest the possibility (high likelihood) of more undetected epistatic interactions (that involve some of our newly generated markers) as a cautionary note.

Our results demonstrate that the phenotypic correlations between foraging-choice behavior and the initiation of foraging (PANKIW and PAGE 2001; PANKIW 2003) are reflected in genetic correlations measured at individual loci (pln1). This can be due to tight linkage or pleiotropic effects of individual QTL, both of which have critical implications for adaptive evolution (MAYNARD SMITH et al. 1985; HAWTHORNE and VIA 2001).

The high recombination rate of the honeybee genome (HUNT and PAGE 1995; BEYE et al. 1999) makes genetic linkage as a basis for genetic correlations less likely in honeybees than in many other organisms. Additionally, the genetic correlation between the rate of behavioral maturation and pollen-hoarding and foraging behavior has persisted through three outcrossing events in our current selection lines, and it also has been found in a previous, independent selection experiment (HELLMICH et al. 1985; CALDERONE and PAGE 1991). The pleiotropic link between the maturation rate and foraging behavior is further corroborated by independent studies of Africanized and European honeybees (PANKIW 2003). Pleiotropy suggests a common proximate basis for pollen hoarding, foraging behavior (HUNT et al. 1995; PAGE et al. 2000), and the rate of behavioral maturation. Genes involved in the perception of brood pheromone (e.g., BRIAND et al. 2002) thus are one class of candidate genes for our QTL because brood pheromone affects foraging initiation (LE CONTE et al. 2001) and foraging choice (PANKIW et al. 1998) of honeybee workers. However, the extensive pleiotropy, together with the demonstrated epistatic interactions, may also be a signature of components of more central cellular networks (OMHOLT et al. 2000), such as the cAMP-dependent second messenger cascade (HUMPHRIES et al. 2003).

Our analysis of 783 linked markers indicated four significant QTL for the rate of behavioral maturation, two in each backcross. These four QTL all differ from each other and they also differ in the size of their effects. Comparison with results from the HBC and earlier mapping experiments (HUNT et al. 1995; PAGE et al. 2000) convinced us to not demote AFF3 and AFF4 from "significant" to "suggestive" QTL on the basis of the bootstrapped significance values. Although AFF3 and AFF4 have lower LOD scores than AFF1 and AFF2, their effect size (and percentage of variation explained) is actually larger. Furthermore, all previously reported honeybee-foraging QTL (even though they had lower LOD scores than any of the AFF-QTL) could be subsequently verified. This makes us also optimistic about the two additional QTL reported in Figure 5. Three of the markers with the strongest effects on AFF were unlinked to any linkage groups in our genetic maps and are therefore not reflected in our current QTL model. However, they may constitute additional QTL in unmapped regions of the genome.

We conclude that there are several sections of the honeybee genome with major influence on the age at onset of foraging. The systematic study of candidate genes in these regions will be greatly facilitated by the results of the honeybee genome project (http://www.hgsc.bcm.tmc.edu/projects/honeybee/). After collection of possible candidates genes, fine-scale mapping, genetic association studies (HIRSCHHORN et al. 2002), and functional analyses (e.g., AMDAM et al. 2003; BEYE et al. 2003) promise further progress toward our mechanistic understanding of the rate of behavioral maturation in honeybees (PHILLIPS 1999; FLINT and MOTT 2001).

As mentioned above, it is not uncommon in experimental crosses between selected lines to find allelic effects that are opposite to the expectation (see, for example, HUNT et al. 1995 for pln1 and PAGE et al. 2000 for pln3). Our newly isolated QTL showed a clear pattern of allelic effects. In both QTL detected in the HBC, the allelic effect was "correct": The allele that promoted fast behavioral maturation was inherited from the high line that matures faster overall. The allelic effects were also consistent with their origin in all three unlinked markers of major effect on AFF. However, for both QTL detected in the LBC, the pattern was the reverse: The allele that promoted fast behavioral maturation was inherited from the "low" line. There are three potential explanations for this phenomenon: epistasis, overdominance, or random fixation of alleles of contrary effect.

We have found evidence for epistasis in this study, as well as in other studies of the high and low pollen-hoarding lines (R. E. PAGE, T. PANKIW, M. BEYE and D. I. NIELSEN, unpublished results). Thus, interactions among loci seem to be the rule, rather than the exception, in the genetic differentiation between the selected honeybee lines studied here. Epistasis can mask additive genetic effects and allow alleles to be maintained in spite of selection against their additive effect (BRODIE 2000; WADE 2002). Thus, epistasis is a potent explanation for our results that the additive effects of alleles of AFF3 and AFF4 seem to contradict their origin. However, on the basis of this observation, it is impossible to rule out the alternative explanation of overdominance (AFF3 and AFF4 heterozygote individuals may be older than either class of homozygotes when initiating foraging) because we had no control over genetic background independently of the three genotypes at each QTL. In the case of overdominance among alleles from the high and the low line, the artificial selection within the low line would not have operated against the alleles of high (that is, fast) effect. The third explanation, random fixation (within a selection line) of alleles whose effects are contrary to the overall direction of selection (facilitated by dominance), is also possible but less likely. Although the population size of our selection lines was not very large (PAGE and FONDRK 1995), the initial selection was strong (PAGE and FONDRK 1995) and both lines underwent three outcrossing events before this experiment.

The second striking overall result from our QTL mapping was the lack of QTL correspondence between the two maps. We did not find any significant effect in homologous regions of the QTL identified in the alternative backcross. Genomic regions that affected AFF in the HBC did not affect AFF in the LBC and vice versa. This finding could likewise be explained by epistatic interactions or within-locus (dominance) interactions but is incompatible with the notion of overdominance. The smaller sample size of the LBC vs. the HBC may also help to explain why HBC QTL were not found in the LBC (but not vice versa). In sum, epistasis seems to be the most likely explanation for the patterns of QTL effects in this study. To demonstrate epistasis and confirm individual QTL, further investigations of the AFF in the two selection lines are needed, involving selected, marker-assisted crosses.

We do not want to exclude the possibility that there are more genetic factors than reported here (BEAVIS 1994) that influence the rate of behavioral maturation. In fact, our results indicate a number of additional regions with possible effects, only two of which we have listed in this article. The restricted sample size, especially in the LBC, has probably led to the omission of QTL with less pronounced effects and inflated the effects of the reported QTL (BEAVIS 1994). Some QTL might also have been missed because our coverage of the honeybee genome is incomplete. Despite the >3700 cM covered in each map, our markers are not grouped in 16 linkage groups, as would be expected on the basis of cytological findings. This disparity indicates gaps in our genetic map. However, independent maps (HUNT and PAGE 1995; PAGE et al. 2000) indicate similar, slightly smaller, map sizes, which suggests that our gaps are small and represent only a minor portion of the honeybee genome.

In conclusion, our selected lines provide an important tool for understanding the rate of behavioral maturation that leads to differences in the age at foraging initiation, a central parameter in the division of labor in social insect colonies. Our experiment supports the view that the substantial genetic variability for this parameter is accessible for identification and characterization. There are several loci with major effects, and nonadditive complexity seems to play a major role. The identified QTL can be regarded as causally responsible for differences in the rate of behavioral maturation, rather than as a correlate of the initiation of foraging (OHASHI et al. 1999; KUCHARSKI and MALESZKA 2002; ROBINSON 2002). While we could exclude direct effects of two candidate genes, this study, together with the honeybee genome project and emerging molecular tools in the honeybee, provides an important basis for the exciting prospect of identifying unexpected or unknown genes that could play a central role in social organization.

AFLP primers used

Primer Sequence E1 AGACTGCGTACCAATTC–AA E2 AGACTGCGTACCAATTC–AC E3 AGACTGCGTACCAATTC–AG E4 AGACTGCGTACCAATTC–CA E5 AGACTGCGTACCAATTC–CC E6 AGACTGCGTACCAATTC–CG E7 AGACTGCGTACCAATTC–TA E8 AGACTGCGTACCAATTC–GG E9 AGACTGCGTACCAATTC–GC EX AGACTGCGTACCAATTC–TG M1 GATGAGTCCTGAGTAAC–AA M2 GATGAGTCCTGAGTAAC–AC M3 GATGAGTCCTGAGTAAC–AG M4 GATGAGTCCTGAGTAAC–AT M5 GATGAGTCCTGAGTAAC–CA M6 GATGAGTCCTGAGTAAC–CC M7 GATGAGTCCTGAGTAAC–CG M8 GATGAGTCCTGAGTAAC–CT M9 GATGAGTCCTGAGTAAC–GA MX GATGAGTCCTGAGTAAC–GC MY GATGAGTCCTGAGTAAC–TA

We thank Merideth Humphries, Greg Hunt, Mindy Nelson, and Mike Niemela for their great experimental support during this study. This research was funded by a Feodor-Lynen Fellowship of the Alexander-von-Humboldt Foundation to O.R. and National Science Foundation grants IBN 0090482 and IBN 0076811 to R.E.P.

AMDAM, G. V., and S. W. OMHOLT, 2003 The hive bee to forager transition in honeybee colonies: the double repressor hypothesis. J. Theor. Biol. 223: 451–464.[CrossRef][Medline]

AMDAM, G. V., Z. L. P. SIMõES, K. R. GUIDUGLI, K. NORBERG and S. W. OMHOLT, 2003 Disruption of vitellogenin gene function in adult honeybees by intra-abdominal injection of double-stranded DNA. BMC Biotechnol. 3: 1 (http://www.biomedcentral.com/1472-6750/3/1[CrossRef][Medline]

ASíNS, M. J., 2002 Present and future of quantitative trait locus analysis in plant breeding. Plant Breed. 121: 281–291.[CrossRef]

BARRON, A. B., D. J. SCHULZ and G. E. ROBINSON, 2002 Octopamine modulates responsiveness to foraging-related stimuli in honey bees (Apis mellifera). J. Comp. Physiol. A 188: 603–610.

BASTEN, C. J., B. S. WEIR, and Z-B. ZENG, 1994 Zmap—a QTL cartographer, pp. 56–66 in Proceedings of the 5th World Congress on Genetics Applied to Livestock Production: Computing Strategies and Software, Vol. 22, edited by C. SMITH, J. S. GAVORA, B. BENKEL, J. CHESNAIS, W. FAIRFULL et al. Organizing Committee, 5th World Congress on Genetics Applied to Livestock Production, Guelph, Ontario, Canada.

BASTEN, C. J., B. S. WEIR and Z-B. ZENG, 2002 QTL Cartographer, Version 1.16: A Reference Manual and Tutorial for QTL Mapping. Department of Statistics, North Carolina State University, Raleigh, NC.

BEAVIS, W. D., 1994 The power and deceit of QTL experiments: lessons from comparative QTL studies. 49th Annual Corn and Sorghum Research Conference. American Seed Trade Association, Washington, DC, pp. 252–268.

BEN-SHAHAR, Y., A. ROBICHON, M. B. SOKOLOWSKI and G. E. ROBINSON, 2002 Influence of gene action across different time scales on behavior. Science 296: 741–744.[Abstract/Free Full Text]

BESHERS, S. N., and J. H. FEWELL, 2001 Models of division of labor in social insects. Annu. Rev. Entomol. 46: 413–440.[CrossRef][Medline]

BEYE, M., G. J. HUNT, R. E. PAGE, JR., M. K. FONDRK, L. GROHMANN et al., 1999 Unusually high recombination rate detected in the sex locus region of the honey bee (Apis mellifera). Genetics 153: 1701–1708.[Abstract/Free Full Text]

BEYE, M., M. HASSELMANN, M. K. FONDRK, R. E. PAGE, JR. and S. W. OMHOLT, 2003 The gene csd is the primary signal for sexual development in the honeybee and encodes an SR-type protein. Cell 114: 419–429.[CrossRef][Medline]

BLOCH, G., D. P. TOMA and G. E. ROBINSON, 2001 Behavioral rhythmicity, age, division of labor and period gene expression in the honeybee brain. J. Biol. Rhythms 16: 444–456.[Abstract]

BLOCH, G., D. L. WHEELER and G. E. ROBINSON, 2002 Endocrine influences on the organization of insect societies, pp. 195–236 in Hormones and Behavior, edited by D. W. PFAFF, A. P. ARNOLD, A. M. ETGEN, S. E. FAHRBACH and R. T. RUBIN. Academic Press, New York.

BRIAND, L., N. SWASDIPAN, C. NESPOULOU, V. BEZIRARD, F. BLON et al., 2002 Characterization of a chemosensory protein (ASP3c) from honeybee (Apis mellifera L.) as a brood pheromone carrier. Eur. J. Biochem. 269: 4586–4596.[Medline]

BRILLET, C., G. E. ROBINSON, R. BUES and Y. LE CONTE, 2002 Racial differences in division of labor in colonies of the honey bee (Apis mellifera). Ethology 108: 115–126.[CrossRef]

BRODIE, III, E. D., 2000 Why evolutionary genetics does not always add up, pp. 3–19 in Epistasis and the Evolutionary Process, edited by J. B.WOLF, E. D. BRODIE III and M. J. WADE. Oxford University Press, New York.

BURKE, J. M., S. TANG, S. J. KNAPP and L. H. RIESEBERG, 2002 Genetic analysis of sunflower domestication. Genetics 161: 1257–1267.[Abstract/Free Full Text]

CALDERONE, N. W., and R. E. PAGE, JR., 1988 Genotypic variability in age polyethism and task specialization in the honeybee, Apis mellifera (Hymenoptera: Apidae). Behav. Ecol. Sociobiol. 22: 17–25.

CALDERONE, N. W., and R. E. PAGE, JR., 1991 Evolutionary genetics of division of labor in colonies of the honey bee (Apis mellifera). Am. Nat. 138: 69–92.

CHURCHILL, G. A., and R. W. DOERGE, 1994 Empirical threshold values for quantitative trait mapping. Genetics 138: 963–971.[Abstract]

FALCONER, D. S., and T. F. C. MACKAY, 1996 Quantitative Genetics, Ed. 4. Addison Wesley Longman, Harlow, UK.

FINCH, C. E., and M. R. ROSE, 1995 Hormones and the physiological architecture of life-history evolution. Q. Rev. Biol. 70: 1–52.[CrossRef][Medline]

FLINT, J., and R. MOTT, 2001 Finding the molecular basis of quantitative traits: successes and pitfalls. Nat. Rev. Genet. 2: 437–445.[CrossRef][Medline]

FRANKEL, W. N., and N. J. SCHORK, 1996 Who's afraid of epistasis? Nat. Genet. 14: 371–373.[CrossRef][Medline]

GIRAY, T., E. GUZMAN-NOVOA, C. W. ARON, B. ZELINSKY, S. E. FAHRBACH et al., 2000 Genetic variation in worker temporal polyethism and colony defensiveness in the honey bee, Apis mellifera. Behav. Ecol. 11: 44–55.[Abstract/Free Full Text]

HAWTHORNE, D. J., and S. VIA, 2001 Genetic linkage of ecological specialization and reproductive isolation in pea aphids. Nature 412: 904–907.[CrossRef][Medline]

HELLMICH, R. L., J. M. KULINCEVIC and W. C. ROTHENBUHLER, 1985 Selection for high and low pollen-hoarding honey bees. J. Hered. 76: 155–158.[Abstract/Free Full Text]

HIRSCHHORN, J. N., K. LOHMUELLER, E. BYRNE and K. A. HIRSCHHORN, 2002 A comprehensive review of genetic association studies. Genet. Med. 4: 45–61.[Medline]

HUANG, Z. Y., and G. E. ROBINSON, 1992 Colony behavioral integration in honey bees: worker-worker interactions mediate plasticity in hormonally regulated division of labor. Proc. Natl. Acad. Sci. USA 89: 11726–11729.[Abstract/Free Full Text]

HUANG, Z. Y., and G. E. ROBINSON, 1996 Regulation of division of labor in honey bees via colony age demography. Behav. Ecol. Sociobiol. 39: 147–158.[CrossRef]

HUMPHRIES, M. A., U. MULLER, M. K. FONDRK and R. E. PAGE, JR., 2003 PKA and PKC content in the honey bee central brain differs in genotypic strains with distinct foraging behavior. J. Comp. Physiol. A 189: 555–562.[CrossRef]

HUNT, G. J., and R. E. PAGE, JR., 1995 Linkage map of the honey bee, Apis mellifera, based on RAPD markers. Genetics 139: 1371–1382.[Abstract]

HUNT, G. J., R. E. PAGE, JR., M. K. FONDRK and C. J. DULLUM, 1995 Major quantitative trait loci affecting honey bee foraging behavior. Genetics 141: 1537–1545.[Abstract]

JANSEN, R.C., 1993 Interval mapping of multiple quantitative trait loci. Genetics 135: 205–211.[Abstract]

KUCHARSKI, R., and R. MALESZKA, 2002 Evaluation of differential gene expression during behavioral development in the honey bee using microarrays and Northern blots. Genome Biol. 3: research 0007.1–0007.9 (http://genomebiology.com/2002/3/2/research/0007).

LAIDLAW, JR., H. H., and R. E. PAGE, JR., 1997 Queen Rearing and Bee Breeding. Wicwas Press, Cheshire, CT.

LANDER, E. S., and D. BOTSTEIN, 1989 Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121: 185–199.[Abstract/Free Full Text]

LECONTE, Y., M. ARESKI and G. E. ROBINSON, 2001 Primer effects of a brood pheromone on honey bee behavioral development. Proc. R. Soc. Lond. B Biol. Sci. 268: 163–168.[Medline]

LINCOLN, S. E., M. J. DALY and E. S. LANDER, 1993 Constructing Genetic Linkage Maps with MAPMAKER/EXP Version 3.0: A Tutorial and Reference Manual. Technical Report, Whitehead Institute for Biomedical Research, Cambridge, MA.

MAYNARD SMITH, J., R. BURIAN, S. KAUFFMAN, P. ALBERCH, J. CAMPELL et al., 1985 Developmental constraints and evolution. Q. Rev. Biol. 60: 265–287.[CrossRef]

MENZEL, R., 1999 Memory dynamics in the honeybee. J. Comp. Physiol. A Sens. Neural Behav. Physiol. 185: 323–340.[CrossRef]

MOUSSEAU, T. A., and D. A. ROFF, 1987 Natural selection and the heritability of fitness components. Heredity 59: 181–197.

OHASHI, K, S. NATORI and T. KUBO, 1999 Expression of amylase and glucose oxidase in the hypopharyngeal gland with an age dependent role change of the worker honeybee (Apis mellifera L.). Eur. J. Biochem. 265: 127–133.[Medline]

OMHOLT, S. W., E. PLAHTE, L. OYEHAUG and K. F. XIANG, 2000 Gene regulatory networks generating the phenomena of additivity, dominance and epistasis. Genetics 155: 969–980.[Abstract/Free Full Text]

PAGE, JR., R. E., and J. ERBER, 2002 Levels of behavioral organization and the evolution of division of labor. Naturwissenschaften 89: 91–106.[CrossRef][Medline]

PAGE, JR., R. E., and M. K. FONDRK, 1995 The effects of colony level selection on the social organization of honey bee (Apis mellifera L.) colonies. Colony level components of pollen hoarding. Behav. Ecol. Sociobiol. 36: 135–144.[CrossRef]

PAGE, JR., R. E., and H. H. LAIDLAW, JR., 1988 Full sisters and super sisters: a terminological paradigm. Anim. Behav. 36: 944–945.[CrossRef]

PAGE, JR., R. E., G. E. ROBINSON, D. S. BRITTON and M. K. FONDRK, 1991 Genotypic variability for rates of behavioral development in worker honey bees (Apis mellifera L.). Behav. Ecol. Sociobiol. 3: 173–180.

PAGE, JR., R. E., G. E. ROBINSON, M. K. FONDRK and M. E. NASR, 1995 Effects of worker genotypic diversity on honey bee colony development and behavior. Behav. Ecol. Sociobiol. 36: 387–396.[CrossRef]

PAGE, JR., R. E., J. ERBER and M. K. FONDRK, 1998 The effect of genotype on response thresholds to sucrose and foraging behavior of honey bees (Apis mellifera L.). J. Comp. Physiol. A 182: 489–500.[CrossRef][Medline]

PAGE, JR., R. E., M. K. FONDRK, G. J. HUNT, E. GUZMáN-NOVOA, M. A. HUMPHRIES et al., 2000 Genetic dissection of honeybee (Apis mellifera L.) foraging behavior. J. Hered. 91: 474–479.[Abstract/Free Full Text]

PAGE, JR., R. E., J. GADAU and M. BEYE, 2002 The emergence of hymenopteran genetics. Genetics 160: 375–379.[Free Full Text]

PANKIW, T., 2003 Directional change in a suite of foraging behaviors in tropical and temperate evolved honey bees (Apis mellifera L.). Behav. Ecol. Sociobiol. 54: 458–464.[CrossRef]

PANKIW, T., and R. E. PAGE, JR., 2001 Genotype and colony environment affect honeybee (Apis mellifera L.) development and foraging behavior. Behav. Ecol. Sociobiol. 51: 87–94.[CrossRef]

PANKIW, T., Z. Y. HUANG and M. L. WINSTON, 1998 Queen mandibular pheromone influences worker honey bee (Apis mellifera) foraging ontogeny and juvenile hormone titers. J. Insect Physiol. 44: 685–692.[CrossRef][Medline]

PANKIW, T., D. R. TARPY and R. E. PAGE, JR., 2002 Genotype and rearing environment affect honeybee perception and foraging behaviour. Anim. Behav. 64: 663–672.[CrossRef]

PHILLIPS, P. C., 1999 From complex traits to complex alleles. Trends Genet. 15: 6–8.[CrossRef][Medline]

ROBINSON, G. E., 2002 Genomics and integrative analyses of division of labor in honey bee colonies. Am. Nat. 160: S160–S172.[CrossRef][Medline]

ROBINSON, G. E., R. E. PAGE, JR., C. STRAMBI and A. STRAMBI, 1989 Hormonal and genetic control of behavioral integration in honey bee colonies. Science 246: 109–112.[Abstract/Free Full Text]

ROBINSON, G. E., R. E. PAGE, JR. and Z. Y. HUANG, 1994 Temporal polyethism in social insects is a developmental process. Anim. Behav. 48: 467–469.[CrossRef]

ROBINSON, G. E., S. E. FAHRBACH and M. L. WINSTON, 1997 Insect societies and molecular biology of social behavior. BioEssays 19: 1099–1108.[CrossRef][Medline]

RüPPELL, O., J. HEINZE and B. HOELLDOBLER, 2001 Complex determination of queen body size in the queen size dimorphic ant Leptothorax rugatulus (Formicidae: Hymenoptera). Heredity 87: 33–40.[CrossRef][Medline]

SANDERSON, A. R., and D. W. HALL, 1948 The cytology of the honey bee, Apis mellifica L. Nature 162: 34–35.[Medline]

SCHEINER, R., R. E. PAGE, JR. and J. ERBER, 2001 Responsiveness to sucrose affects tactile and olfactory learning in preforaging honey bees of two genetic strains. Behav. Brain Res. 120: 67–73.[CrossRef][Medline]

SCHULZ, D. J., and G. E. ROBINSON, 1999 Biogenic amines and division of labor in honey bees: age related changes in the mushroom bodies and behaviorally related changes in the antennal lobes. J. Comp. Physiol. A 184: 481–488.[CrossRef][Medline]

SCHULZ, D. J., and G. E. ROBINSON, 2001 Octopamine influences division of labor in honey bee colonies. J. Comp. Physiol. A 187: 53–61.[CrossRef][Medline]

SCHULZ, D. J., Z. Y. HUANG and G. E. ROBINSON, 1998 Effects of colony food shortage on behavioral development in honey bees. Behav. Ecol. Sociobiol. 42: 295–303.[CrossRef]

SCHULZ, D. J., J. P. SULLIVAN and G. E. ROBINSON, 2002 Juvenile hormone and octopamine in the regulation of division of labor in honey bee colonies. Horm. Behav. 42: 222–231.[CrossRef][Medline]

SULLIVAN, J. P., O. JASSIM, S. E. FAHRBACH and G. E. ROBINSON, 2000 Juvenile hormone paces behavioral development in the adult worker honey bee. Horm. Behav. 37: 1–14.[CrossRef][Medline]

TEMPLETON, A. R., 2000 Epistasis and complex traits, pp. 41–56 in Epistasis and the Evolutionary Process, edited by J. B. WOLF, E. D. BRODIE III and M. J. WADE. Oxford University Press, New York.

TOMA, D. P., G. BLOCH, D. MOORE and G. E. ROBINSON, 2000 Changes in period mRNA levels in the brain and division of labor in honey bee colonies. Proc. Natl. Acad. Sci. USA 97: 6914–6919.[Abstract/Free Full Text]

VAN OOIJEN, J. W., M. P. BOER, R. C. JANSEN and C. MALIEPAARD, 2002 MapQTL 4.0, Software for the Calculation of QTL Positions on Genetic Maps. Plant Research International, Wageningen, The Netherlands.

WADE, M. J., 2002 A gene's eye view of epistasis, selection and speciation. J. Evol. Biol. 15: 337–346.[CrossRef]

WINSTON, M. L., 1987 The Biology of the Honey bee. Harvard University Press, Cambridge, MA.

WOLF, J. B., 2000 Indirect genetic effects and gene interactions, pp. 158–176 in Epistasis and the Evolutionary Process, edited by J. B. WOLF, E. D. BRODIE III and M. J. WADE, Oxford University Press, New York.

This article has been cited by other articles:

				O. Rueppell, S. B. C. Chandra, T. Pankiw, M. K. Fondrk, M. Beye, G. Hunt, and R. E. Page The Genetic Architecture of Sucrose Responsiveness in the Honeybee (Apis mellifera L.) Genetics, January 1, 2006; 172(1): 243 - 251. [Abstract] [Full Text] [PDF]

International Access Link

Online ISS 1091-6490
Copyright 2008 by the Genetics Society of America
phone: 412-268-1812   fax: 412-268-1813   email: genetics-gsa{at}andrew.cmu.edu