First large-scale genomic prediction in the honey bee.

Genomic selection has increased genetic gain in several livestock species, but due to the complicated genetics and reproduction biology not yet in honey bees. Recently, 2970 queens were genotyped to gather a reference population. For the application of genomic selection in honey bees, this study analyzes the accuracy and bias of pedigree-based and genomic breeding values for honey yield, three workability traits, and two traits for resistance against the parasite Varroa destructor. For breeding value estimation, we use a honey bee-specific model with maternal and direct effects, to account for the contributions of the workers and the queen of a colony to the phenotypes. We conducted a validation for the last generation and a five-fold cross-validation. In the validation for the last generation, the accuracy of pedigree-based estimated breeding values was 0.12 for honey yield, and ranged from 0.42 to 0.61 for the workability traits. The inclusion of genomic marker data improved these accuracies to 0.23 for honey yield, and a range from 0.44 to 0.65 for the workability traits. The inclusion of genomic data did not improve the accuracy of the disease-related traits. Traits with high heritability for maternal effects compared to the heritability for direct effects showed the most promising results. For all traits except the Varroa resistance traits, the bias with genomic methods was on a similar level compared to the bias with pedigree-based BLUP. The results show that genomic selection can successfully be applied to honey bees.

Consequences of incorrect genetic parameter estimates for single-trait and multi-trait genetic evaluations in honeybees.

Genetic and residual variances of traits are important input parameters for best linear unbiased prediction (BLUP) breeding value estimation. In honeybees, estimates of these variances are often associated with large standard errors, entailing a risk to perform genetic evaluations under wrong premises. The consequences hereof have not been sufficiently studied. In particular, there are no adequate investigations on this topic accounting for multi-trait selection or genetic peculiarities of the honeybee. We performed simulation studies and explored the consequences of selection for honeybee populations with a broad range of true and assumed genetic parameters. We found that in single-trait evaluations, the response to selection was barely compromised by assuming erroneous parameters, so that reductions in genetic progress after 20 years never exceeded 21%. Phenotypic selection appeared inferior to BLUP selection, particularly under low heritabilities. Parameter choices for genetic evaluation had great effects on inbreeding development. By wrongly assuming high heritabilities, inbreeding rates were reduced by up to 74%. When parallel selection was performed for two traits, the right choice of genetic parameters appeared considerably more crucial as several incorrect premises yielded inadvertent negative selection for one of the traits. This phenomenon occurred in multiple constellations in which the selection traits expressed a negative genetic correlation. It was not reflected in the estimated breeding values. Our results indicate that breeding efforts heavily rely on detailed knowledge on genetic parameters, particularly when multi-trait selection is performed. Thus, considerable effort should be invested into precise parameter estimations.

Complex population structure and haplotype patterns in the Western European honey bee from sequencing a large panel of haploid drones.

Honey bee subspecies originate from specific geographical areas in Africa, Europe and the Middle East, and beekeepers interested in specific phenotypes have imported genetic material to regions outside of the bees' original range for use either in pure lines or controlled crosses. Moreover, imported drones are present in the environment and mate naturally with queens from the local subspecies. The resulting admixture complicates population genetics analyses, and population stratification can be a major problem for association studies. To better understand Western European honey bee populations, we produced a whole genome sequence and single nucleotide polymorphism (SNP) genotype data set from 870 haploid drones and demonstrate its utility for the identification of nine genetic backgrounds and various degrees of admixture in a subset of 629 samples. Five backgrounds identified correspond to subspecies, two to isolated populations on islands and two to managed populations. We also highlight several large haplotype blocks, some of which coincide with the position of centromeres. The largest is 3.6 Mb long and represents 21% of chromosome 11, with two major haplotypes corresponding to the two dominant genetic backgrounds identified. This large naturally phased data set is available as a single vcf file that can now serve as a reference for subsequent populations genomics studies in the honey bee, such as (i) selecting individuals of verified homogeneous genetic backgrounds as references, (ii) imputing genotypes from a lower-density data set generated by an SNP-chip or by low-pass sequencing, or (iii) selecting SNPs compatible with the requirements of genotyping chips.

Semi-automatic detection of honeybee brood hygiene-an example of artificial learning to facilitate ethological studies on social insects.

Machine-learning techniques are shifting the boundaries of feasibility in many fields of ethological research. Here, we describe an application of machine learning to the detection/measurement of hygienic behaviour, an important breeding trait in the honey bee (Apis mellifera). Hygienic worker bees are able to detect and destroy diseased brood, thereby reducing the reproduction of economically important pathogens and parasites such as the Varroa mite (Varroa destructor). Video observation of this behaviour on infested combs has many advantages over other methods of measurement, but analysing the recorded material is extremely time-consuming. We approached this problem by combining automatic tracking of bees in the video recordings, extracting relevant features, and training a multi-layer discriminator on positive and negative examples of the behaviour of interest. Including expert knowledge into the design of the features lead to an efficient model for identifying the uninteresting parts of the video which can be safely skipped. This algorithm was then used to semiautomatically identify individual worker bees involved in the behaviour. Application of the machine-learning method allowed to save 70% of the time required for manual analysis, and substantially increased the number of cell openings correctly identified. It thereby turns video-observation of individual cell opening events into an economically competitive method for selecting potentially resistant bees. This method presents an example of how machine learning can be used to boost ethological research, and how it can generate new knowledge by explaining the learned decision rule in form of meaningful parameters.

Influence of model selection and data structure on the estimation of genetic parameters in honeybee populations.

Estimating genetic parameters of quantitative traits is a prerequisite for animal breeding. In honeybees, the genetic variance separates into queen and worker effects. However, under data paucity, parameter estimations that account for this peculiarity often yield implausible results. Consequently, simplified models that attribute all genetic contributions to either the queen (queen model) or the workers (worker model) are often used to estimate variance components in honeybees. However, the causes for estimations with the complete model (colony model) to fail and the consequences of simplified models for variance estimates are little understood. We newly developed the necessary theory to compare parameter estimates that were achieved by the colony model with those of the queen and worker models. Furthermore, we performed computer simulations to quantify the influence of model choice, estimation algorithm, true genetic parameters, rates of controlled mating, apiary sizes, and phenotype data completeness on the success of genetic parameter estimations. We found that successful estimations with the colony model were only possible if at least some of the queens mated controlled on mating stations. In that case, estimates were largely unbiased if more than 20% of the colonies had phenotype records. The simplified queen and worker models proved more stable and yielded plausible parameter estimates for almost all settings. Results obtained from these models were unbiased when mating was uncontrolled, but with controlled mating, the simplified models consistently overestimated heritabilities. This study elucidates the requirements for variance component estimation in honeybees and provides the theoretical groundwork for simplified honeybee models.

Suitability of drone olfactory sensitivity as a selection trait for Varroa-resistance in honeybees.

The most effective strategy against brood diseases, such as those stemming from infestation by the mite Varroa destructor, is the early detection and removal of sick brood. Recent findings suggest that genes associated with worker bee olfactory perception play a central role in Varroa-sensitive hygiene (VSH). In this study, the odour sensitivity of Apis mellifera drones was examined through proboscis extension response (PER) conditioning. Individuals sensitive/insensitive to the two Varroa-parasitised-brood odours (extract-low and extract-high) were used for breeding. Twenty-one queens from a VSH-selected line (SelQ) and nineteen queens from a nonselected line (ConQ) were single-drone-inseminated with sperm from drones that showed either sensitivity (SenD+) or insensitivity (SenD-) to the two extracts. Individual VSH behaviour in a total of 5072 offspring of these combinations (SelQ × SenD+, SelQ × SenD-, ConQ × SenD+, ConQ × SenD-) was subsequently observed in a specially designed observation unit with infrared light. The results from the video observation were also separately examined, considering the genetic origin (VSH-selected or nonselected line) of the participating queens and drones. While the drone PER conditioning results were not significantly reflected in the VSH results of the respective offspring, the genetic origin of the participating queens/drones was crucial for VSH manifestation.

Strength surpasses relatedness-queen larva selection in honeybees.

Nepotism was initially theoretically predicted and sometimes found to trigger the selection of specific larvae to be reared as queens in the honeybee Apis mellifera. Although the importance of selecting the next queen for a colony indicates that it should not occur at random, nepotism is increasingly considered unlikely in eusocial insect societies. Different prenatal maternal supplies of embryos have been found to impact fitness in many other species and therefore could be a possible trigger underlying the likelihood of being raised as a queen. We offered related or unrelated larvae from six colonies originating from eggs of different weights for emergency queen rearing in queenless units with worker bees from these six colonies. We showed that nurses did not significantly prefer related larvae during queen rearing, which confirms the theory that different relatedness-driven kin preferences within a colony cannot be converted into a colony-level decision. However, we found that larvae originating from heavier eggs were significantly preferred for queen breeding. Studies on other species have shown that superior maternal supply is important for later reproductive success. However, we did observe tendencies in the expected direction (e.g., queens that hatched from heavier eggs had both more ovarioles and a shorter preoviposition period). Nevertheless, our data do not allow for a significant conclusion that the selection of larvae from heavy eggs truly offers fitness advantages.

Simulation studies to optimize genomic selection in honey bees.

BACKGROUND: With the completion of a single nucleotide polymorphism (SNP) chip for honey bees, the technical basis of genomic selection is laid. However, for its application in practice, methods to estimate genomic breeding values need to be adapted to the specificities of the genetics and breeding infrastructure of this species. Drone-producing queens (DPQ) are used for mating control, and usually, they head non-phenotyped colonies that will be placed on mating stations. Breeding queens (BQ) head colonies that are intended to be phenotyped and used to produce new queens. Our aim was to evaluate different breeding program designs for the initiation of genomic selection in honey bees. METHODS: Stochastic simulations were conducted to evaluate the quality of the estimated breeding values. We developed a variation of the genomic relationship matrix to include genotypes of DPQ and tested different sizes of the reference population. The results were used to estimate genetic gain in the initial selection cycle of a genomic breeding program. This program was run over six years, and different numbers of genotyped queens per year were considered. Resources could be allocated to increase the reference population, or to perform genomic preselection of BQ and/or DPQ. RESULTS: Including the genotypes of 5000 phenotyped BQ increased the accuracy of predictions of breeding values by up to 173%, depending on the size of the reference population and the trait considered. To initiate a breeding program, genotyping a minimum number of 1000 queens per year is required. In this case, genetic gain was highest when genomic preselection of DPQ was coupled with the genotyping of 10-20% of the phenotyped BQ. For maximum genetic gain per used genotype, more than 2500 genotyped queens per year and preselection of all BQ and DPQ are required. CONCLUSIONS: This study shows that the first priority in a breeding program is to genotype phenotyped BQ to obtain a sufficiently large reference population, which allows successful genomic preselection of queens. To maximize genetic gain, DPQ should be preselected, and their genotypes included in the genomic relationship matrix. We suggest, that the developed methods for genomic prediction are suitable for implementation in genomic honey bee breeding programs.

Short-term effects of controlled mating and selection on the genetic variance of honeybee populations.

Directional selection in a population yields reduced genetic variance due to the Bulmer effect. While this effect has been thoroughly investigated in mammals, it is poorly studied in social insects with biological peculiarities such as haplo-diploidy or the collective expression of traits. In addition to the natural adaptation to climate change, parasites, and pesticides, honeybees increasingly experience artificial selection pressure through modern breeding programs. Besides selection, many honeybee breeding schemes introduce controlled mating. We investigated which individual effects selection and controlled mating have on genetic variance. We derived formulas to describe short-term changes of genetic variance in honeybee populations and conducted computer simulations to confirm them. Thereby, we found that the changes in genetic variance depend on whether the variance is measured between queens (inheritance criterion), worker groups (selection criterion), or both (performance criterion). All three criteria showed reduced genetic variance under selection. In the selection and performance criteria, our formulas and simulations showed an increased genetic variance through controlled mating. This newly described effect counterbalanced and occasionally outweighed the Bulmer effect. It could not be observed in the inheritance criterion. A good understanding of the different notions of genetic variance in honeybees, therefore, appears crucial to interpreting population parameters correctly.

A theoretical derivation of response to selection with and without controlled mating in honeybees.

BACKGROUND: In recent years, the breeding of honeybees has gained significant scientific interest, and numerous theoretical and practical improvements have been made regarding the collection and processing of their performance data. It is now known that the selection of high-quality drone material is crucial for mid to long-term breeding success. However, there has been no conclusive mathematical theory to explain these findings. METHODS: We derived mathematical formulas to describe the response to selection of a breeding population and an unselected passive population of honeybees that benefits indirectly from genetic improvement in the breeding population via migration. This was done under the assumption of either controlled or uncontrolled mating of queens in the breeding population. RESULTS: Our model equations confirm what has been observed in simulation studies. In particular, we have proven that the breeding population and the passive population will show parallel genetic gain after some years and we were able to assess the responses to selection for different breeding strategies. Thus, we confirmed the crucial importance of controlled mating for successful honeybee breeding. When compared with data from simulation studies, the derived formulas showed high coefficients of determination [Formula: see text] in cases where many passive queens had dams from the breeding population. For self-sufficient passive populations, the coefficients of determination were lower ([Formula: see text]) if the breeding population was under controlled mating. This can be explained by the limited simulated time-frame and lower convergence rates. CONCLUSION: The presented theoretical derivations allow extrapolation of honeybee-specific simulation results for breeding programs to a wide range of population parameters. Furthermore, they provide general insights into the genetic dynamics of interdependent populations, not only for honeybees but also in a broader context.

Substantial Genetic Progress in the International Apis mellifera carnica Population Since the Implementation of Genetic Evaluation.

The Apis mellifera carnica subspecies of the honeybee has long been praised for its gentleness and good honey yield before systematic breeding efforts began in the early 20th century. However, before the introduction of modern techniques of genetic evaluation (best linear unbiased prediction, BLUP) and a computerized data management in the mid 1990s, genetic progress was slow. Here, the results of the official breeding value estimation in BeeBreed.eu are analyzed to characterize breeding progress and inbreeding. From about the year 2000 onward, the genetic progression accelerated and resulted in a considerable gain in honey yield and desirable properties without increased inbreeding coefficients. The prognostic quality of breeding values is demonstrated by a retrospective analysis. The success of A. m. carnica breeding shows the potential of BLUP-based breeding values and serves as an example for a large-scale breeding program.

Individual-Level Comparisons of Honey Bee (Hymenoptera: Apoidea) Hygienic Behavior Towards Brood Infested with Varroa destructor (Parasitiformes: Varroidae) or Tropilaelaps mercedesae (Mesostigmata: Laelapidae).

The mites Varroa destructor Anderson and Trueman and Tropilaelaps mercedesae Anderson and Morgan are both serious threats to the Apis mellifera beekeeping industry. A trait frequently used in selection programs for V. destructor resistance is hygienic behavior, the selective removal of diseased/damaged brood. Here, we measured the level of association of the expression of hygienic behavior against both mites in A. mellifera, by observing whether the same individual bees would carry out the opening and removal of brood infested by the two parasites. The groups of bees showing these behaviors on cells artificially infested by either parasite showed a large overlap, making it appear likely that the two traits are at least closely coupled. Therefore, breeding for V. destructor resistance based on hygienic behavior could prepare A. mellifera populations for dealing with Tropilaelaps sp. mites, and vice versa. Using the same bioassay, we also compared the hygienic behavior of A. mellifera towards T. mercedesae to that of the Asiatic honey bee, Apis cerana. A. cerana workers eliminated a greater proportion of infested cells, which may in part explain the resistance of this bee to Tropilaelaps and the observation that Tropilaelaps reproduction on brood of this species is extremely rare.

Tool for genomic selection and breeding to evolutionary adaptation: Development of a 100K single nucleotide polymorphism array for the honey bee.

High-throughput high-density genotyping arrays continue to be a fast, accurate, and cost-effective method for genotyping thousands of polymorphisms in high numbers of individuals. Here, we have developed a new high-density SNP genotyping array (103,270 SNPs) for honey bees, one of the most ecologically and economically important pollinators worldwide. SNPs were detected by conducting whole-genome resequencing of 61 honey bee drones (haploid males) from throughout Europe. Selection of SNPs for the chip was done in multiple steps using several criteria. The majority of SNPs were selected based on their location within known candidate regions or genes underlying a range of honey bee traits, including hygienic behavior against pathogens, foraging, and subspecies. Additionally, markers from a GWAS of hygienic behavior against the major honey bee parasite Varroa destructor were brought over. The chip also includes SNPs associated with each of three major breeding objectives-honey yield, gentleness, and Varroa resistance. We validated the chip and make recommendations for its use by determining error rates in repeat genotypings, examining the genotyping performance of different tissues, and by testing how well different sample types represent the queen's genotype. The latter is a key test because it is highly beneficial to be able to determine the queen's genotype by nonlethal means. The array is now publicly available and we suggest it will be a useful tool in genomic selection and honey bee breeding, as well as for GWAS of different traits, and for population genomic, adaptation, and conservation questions.

Long-Term Evaluation of Breeding Scheme Alternatives for Endangered Honeybee Subspecies.

Modern breeding structures are emerging for European honeybee populations. However, while genetic evaluations of honeybees are becoming increasingly well understood, little is known about how selection decisions shape the populations' genetic structures. We performed simulations evaluating 100 different selection schemes, defined by selection rates for dams and sires, in populations of 200, 500, or 1000 colonies per year and considering four different quantitative traits, reflecting different genetic parameters and numbers of influential loci. Focusing on sustainability, we evaluated genetic progress over 100 years and related it to inbreeding developments. While all populations allowed for sustainable breeding with generational inbreeding rates below 1% per generation, optimal selection rates differed and sustainable selection was harder to achieve in smaller populations and for stronger negative correlations of maternal and direct effects in the selection trait. In small populations, a third or a fourth of all candidate queens should be selected as dams, whereas this number declined to a sixth for larger population sizes. Furthermore, our simulations indicated that, particularly in small populations, as many sires as possible should be provided. We conclude that carefully applied breeding provides good prospects for currently endangered honeybee subspecies, since sustainable genetic progress improves their attractiveness to beekeepers.

The importance of controlled mating in honeybee breeding.

BACKGROUND: Controlled mating procedures are widely accepted as a key aspect of successful breeding in almost all animal species. In honeybees, however, controlled mating is hard to achieve. Therefore, there have been several attempts to breed honeybees using free-mated queens. In such breeding schemes, selection occurs only on the maternal path since the drone sires are random samples of the population. The success rates of breeding approaches without controlled mating have so far not been investigated on a theoretical or simulation-based level. METHODS: Stochastic simulation studies were carried out to examine the chances of success in honeybee breeding with and without controlled mating. We investigated the influence of different sizes of breeding populations (500, 1000, 2000 colonies per year) and unselected passive populations (0, 500, 1000, 2000, infinitely many colonies per year) on selection for a maternally (queen) and directly (worker group) influenced trait with moderate ([Formula: see text]) or strong ([Formula: see text]) negative correlation between the two effects. The simulations described 20 years of selection. RESULTS: Our simulations showed a reduction of breeding success between 47 and 99% if mating was not controlled. In the most drastic cases, practically no genetic gain could be generated without controlled mating. We observed that in the trade-off between selection for direct or maternal effects, the absence of mating control leads to a shift in favor of maternal effects. Moreover, we describe the implications of different breeding strategies on the unselected passive population that benefits only indirectly via the transfer of queens or drones from the breeding population. We show that genetic gain in the passive population develops parallel to that of the breeding population. However, we found a genetic lag that became significantly smaller as more breeding queens served as dams of queens in the passive population. CONCLUSIONS: We conclude that even when unwanted admixture of subspecies can be excluded in natural matings, controlled mating is imperative for successful breeding efforts. This is especially highlighted by the strong positive impact that controlled mating in the breeding population has on the unselected passive population.

Comparison of infinitesimal and finite locus models for long-term breeding simulations with direct and maternal effects at the example of honeybees.

Stochastic simulation studies of animal breeding have mostly relied on either the infinitesimal genetic model or finite polygenic models. In this study, we investigated the long-term effects of the chosen model on honeybee breeding schemes. We implemented the infinitesimal model, as well as finite locus models, with 200 and 400 gene loci and simulated populations of 300 and 1000 colonies per year over the course of 100 years. The selection was of a directly and maternally influenced trait with maternal heritability of [Formula: see text], direct heritability of [Formula: see text], and a negative correlation between the effects of rmd = - 0.18. Another set of simulations was run with parameters [Formula: see text], [Formula: see text], and rmd = - 0.53. All models showed similar behavior for the first 20 years. Throughout the study, we observed a higher genetic gain in the direct than in the maternal effects and a smaller gain with a stronger negative covariance. In the long-term, however, only the infinitesimal model predicted sustainable linear genetic progress, while the finite locus models showed sublinear behavior and, after 100 years, only reached between 58% and 62% of the mean breeding values in the infinitesimal model. While the infinitesimal model suggested a reduction of genetic variance by 33% to 49% after 100 years, the finite locus models saw a more drastic loss of 76% to 92%. When designing sustainable breeding strategies, one should, therefore, not blindly trust the infinitesimal model as the predictions may be overly optimistic. Instead, the more conservative choice of the finite locus model should be favored.

Recognition of mite-infested brood by honeybee (Apis mellifera) workers may involve thermal sensing.

Hygienic behavior, i.e. the removal of diseased or damaged brood by worker honey bees (Apis mellifera), is seen as one of the principal behavioral elements of this species' social immunity. Identification of the stimuli that trigger it would be helpful in searching for biochemical and molecular markers of this important breeding trait. While many studies at the genomic, transcriptomic, and behavioral level have pointed to the implication of chemical cues, we here hypothesized that thermal cues are alternatively/additionally involved. To test this hypothesis, we first measured whether infestation by the mite Varroa destructor (a condition known to induce hygienic behavior) leads to a thermal gradient between affected and unaffected brood. We found that infested brood cells were between 0.03 and 0.19 °C warmer than uninfested controls. Next, we tested whether artificially heating an area of a brood comb would increase the removal of infested or uninfested brood as compared to an unheated control area, and found that this was not the case. Finally, we investigated whether the heating of individual brood cells, as opposed to comb areas, would influence brood removal from cells adjacent to the heated one. This was the case for uninfested, though not for infested cells. We conclude that infestation by V. destructor leads to a heating of brood cells that should be perceivable by bees, and that small-scale temperature gradients can influence brood removal. This makes it appear possible that thermal cues play a role in triggering hygienic behavior of honey bees directed at varroa-infested larvae/pupae, although our results are insufficient to prove such an involvement.

Behavioral and molecular studies of quantitative differences in hygienic behavior in honeybees.

BACKGROUND: Hygienic behavior (HB) enables honeybees to tolerate parasites, including infection with the parasitic mite Varroa destructor, and it is a well-known example of a quantitative genetic trait. The understanding of the molecular processes underpinning the quantitative differences in this behavior remains limited. RESULTS: We performed gene expression studies in worker bees that displayed quantitative genetic differences in HB. We established a high and low genetic source of HB performance and studied the engagements into HB of single worker bees under the same environmental conditions. We found that the percentage of worker bees that engaged in a hygienic behavioral task tripled in the high versus low HB sources, thus suggesting that genetic differences may mediate differences in stimulated states to perform HB. We found 501 differently expressed genes (DEGs) in the brains of hygienic and non-hygienic performing workers in the high HB source bees, and 342 DEGs in the brains of hygienic performing worker bees, relative to the gene expression in non-hygienic worker bees from the low HB source group. "Cell surface receptor ligand signal transduction" in the high and "negative regulation of cell communication" in the low HB source were overrepresented molecular processes, suggesting that these molecular processes in the brain may play a role in the regulation of quantitative differences in HB. Moreover, only 21 HB-associated DEGs were common between the high and low HB sources. CONCLUSIONS: The better HB colony performance is primarily achieved by a high number of bees engaging in the hygienic tasks that associate with distinct molecular processes in the brain. We propose that different gene products and pathways may mediate the quantitative genetic differences of HB.

Secondary biomarkers of insecticide-induced stress of honey bee colonies and their relevance for overwintering strength.

The evaluation of pesticide side-effects on honeybees is hampered by a lack of colony-level bioassays that not only are sensitive to physiological changes, but also allow predictions about the consequences of exposure for longer-term colony productivity and survival. Here we measured 28 biometrical, biochemical and behavioural indicators in a field study with 63 colonies and 3 apiaries. Colonies were stressed in early summer by feeding them for five days with either the carbamate growth regulator fenoxycarb or the neurotoxic neonicotinoid imidacloprid, or left untreated. Candidate stress indicators were measured 8-64 days later. We determined which of the indicators were influenced by the treatments, and which could be used as predictors in regression analyses of overwintering strength. Among the indicators influenced by fenoxycarb were the amount of brood in colonies as well as the learning performance and 24h-memory of bees, and the concentration of the brood food component 10HDA in head extracts. Imidacloprid significantly affected honey production, total number of bees and activity of the immune-related enzyme phenoloxidase in forager bee extracts. Indicators predictive of overwintering strength but unrelated to insecticide feeding included vitellogenin titer and glucose oxidase-activity in haemolymph/whole body-extracts of hive bees. Apart from variables that were themselves components of colony strength (numbers of bees/brood cells), the only indicator that was both influenced by an insecticide and predictive of overwintering strength was the concentration of 10HDA in worker bee heads. Our results show that physiological and biochemical bioassays can be used to study effects of insecticides at the colony level and assess the vitality of bee colonies. At the same time, most bioassays evaluated here appear of limited use for predicting pesticide effects on colony overwintering strength, because those that were sensitive to the insecticides were not identical with those that were predictive of colony overwintering. Our study therefore illustrates the difficulties involved in evaluating the economic/ecological significance of pesticide-induced stress in honey bee field studies.