Overview of Feral and Managed Honeybees in Australia

Distribution, Abundance, Extent of Interactions with Native Biota, Evidence of Impacts and Future Research
David C Paton, Department of Zoology, The University of Adelaide
for Australian Nature Conservation Agency
Environment Australia, May 1996
ISBN 0 6422 1381 X

5. Future Research


Effective management of honeybees in natural systems will eventually depend on accumulating sufficient information on:

  1. the biology of most of the native taxa that now interact with honeybees and whether those taxa suffer or benefit from the interaction;
  2. the ecology of feral and managed populations of honeybees; and
  3. efficient and effective ways of removing feral colonies of honeybees from areas.

Future research needs to address each of these.

Modern ecological approaches demand manipulative experiments to test for interactions between taxa and to assess the benefits of various management actions. Wapshere (1988) outlines some of the manipulative experiments needed to test hypotheses concerning the effects of honeybees on native flora and fauna. Manipulative experiments, however, cannot be designed or executed without some knowledge of the systems in the first place and so descriptive studies will often need to precede experimental studies. Descriptive studies are particularly important in that they allow appropriate temporal and spatial scales to be determined for experiments as well as providing information on the inherent variability within these systems. Knowledge of that variability allows adequate levels of replication to be set for experimental studies.

Examples of the types of descriptive and experimental studies that are needed are given below and include an outline of the methods and a discussion of the factors that need to be considered when designing field experiments. For convenience these studies are grouped under five headings:

  1. Research on interactions between honeybees, flower-visiting fauna and flora.
  2. Studies of hollow use by native fauna and honeybees.
  3. Studies on the population dynamics of feral honeybees to identify limiting factors.
  4. Research on the efficacy of various techniques to control or eradicate feral honeybees.
  5. Studies on patterns of floral resource production and use of these resources by commercial beekeepers.

Research on interactions between honeybees, flowervisiting fauna and flora

Selecting taxa for study

Several criteria could be used to select taxa for study. The first and foremost criterion should be that the taxon interacts with honeybees. Given that a great diversity of taxa interact with honeybees then a representative range of plants and animals should be selected for study, since any one or all of these taxa may be affected, and/or affected in different ways. Paton (1993) suggests a further refinement. Specific taxa should be selected based on the magnitude of the interactions that are taking place, the magnitude of an interaction being judged by the frequency with which honeybees visit flowers and/or by the proportion or quantities of floral resources that are being consumed by honeybees (eg tables 6,7). When a large share of the floral resources of a plant are being consumed by honeybees a significant impact on either the plants or native fauna is more likely than when only a small proportion of the resources are being consumed. This approach to selecting taxa for study is different to that promoted by most conservation agencies where the focus is on endangered, vulnerable or rare taxa irrespective of the level of interaction. Thus the first stage involves identifying those plants that are being heavily exploited by honeybees. Methods for doing this are outlined in Paton (1990) and are not given here.

Certain other features can also be used to select from this list those taxa that are more likely to be affected by honeybees, The following additional criteria are likely to make certain plants and animals more sensitive to perturbations from honeybees. Plants that are obligate outcrossers, have large floral displays and are widely spaced are more likely to be affected detrimentally by honeybees than those that are self-compatible, have small floral displays and occur in dense aggregations. This is because individual honeybees tend to forage in restricted areas and are less likely to move between plants when the floral displays and the distances between plants are large. Plants that require particular animals to operate the pollination mechanism may also be more susceptible than those that have less specialised flowers that can be pollinated by a range of animals (including honeybees). At present only limited information is available on the reproductive biology of most of the native plants that interact with honeybees. Whether any of these plants depend on particular native fauna for pollination is also often not known.

Selecting appropriate fauna for study is more complex than that for plants because there are usually several to many species of native animals visiting each of the plant species that honeybees might be using extensively. Competition for floral resources, however, is likely to be more severe for native animals that are larger rather than smaller than honeybees. Larger species have greater demands for food resources than smaller species (all other things being equal) and so are more likely to be affected detrimentally if resources at flowers are more heavily cropped. Thus vertebrates may be more sensitive than invertebrates to the foraging activity of honeybees. Other factors that may be important include the time of the day when an animal forages and how this relates to resource availability at flowers, and whether or not the animal specialises on only one or a few key plants and whether these are used extensively by honeybees.

Wapshere (1988) suggests that native bees of the genus Trigona should be studied since these bees are social and hence more similar behaviourally to honeybees than other native fauna. However Trigona are usually much smaller than honeybees and so may not overlap extensively in foraging niche. Others would argue that the largest (>I 0 mm in length) native bees (eg species of Lestis) are more likely to be affected by losses of food resources to honeybees since their food requirements are greater than smaller native bees, or the short-tongued bees because access to nectar is more restricted for them. Previous workers, however, have selected species of native bee that are relatively well known and whose reproductive performance can be measured (eg species of Exoneura; Sugden and Pyke 1991; Schwartz et at.unpubl.) yet these species may not have overlapped extensively in foraging niche with honeybees. A large body of data shows that a wide variety of taxa including birds, native bees and other invertebrates all interact significantly with honeybees and any one of these could be adversely affected by them. A range of plants and animals that interact with honeybees should. be selected for study.

Studies on the plants

For the plants involved in significant interactions with honeybees the first suite of studies should document the plant's reproductive performance. This should consist of:

  1. establishing the breeding system of the plant;
  2. measuring the current rate and variability in seed production both across the flowering season and between individual plants of the same species; and
  3. determining if seed production is limited by pollination.

These basic measurements are important for establishing sample sizes and levels of replication in subsequent experiments and for providing direction for the next stage of the research. For example, if the conversion rate of flowers to fruits is highly variable then more replicates of each of the treatments may be required and a larger number of individual plants may need to be treated within each replicate to provide adequate statistical power to properly test the effects of certain factors (ie honeybees).

To a large extent if the reproductive performance of a plant is not limited by pollination then concerns that honeybees are having a significant detrimental effect on this plant are diminished (at least with respect to the quantity of seeds being produced). However, depending on the frequency with which native fauna visit flowers, these plants may now depend on honeybees for seed production. Some initial assessment of the likelihood of this can be made by calculating the frequency with which native fauna visit flowers. If their visitation rates are low then the plant's dependence on honeybees for pollination may be high. For those plants where seed production is limited by the amount of pollination then further examination of the interactions between this plant, honeybees and native pollinators is warranted. This might involve recording the foraging behaviour of different floral visitors and assessing their value as pollinators. For example, observations might reveal that honeybees only visit male-phase flowers and rob flowers of nectar or pollen without effecting pollination or rarely move between plants and so differ from native pollinators in the pollination services they provide.

In both cases the only way to establish the role that honeybees are playing in their pollination is to exclude honeybees from plots and record seed production in areas with and without honeybees.

Eventually studies that measure the `quality' (genetic diversity) of seed crops may need to be implemented, but the technology for this and value of doing this depends on more detailed knowledge of the plant's pollination biology.

Studies on native fauna that visit flowers

A good understanding of the basic ecology of floral visitors is required before impacts of honeybees can be assessed. Except for some honeyeaters and a few species of native bees, these basic ecological data are lacking. Many species of invertebrates are still to be formally described, let alone studied ecologically. For example, there are an estimated 3000 species of native bees in Australia (Michener 1970; T Houston pers. comm.) of which less than 2000 have been described. Simple information on the distribution and abundance of most species of flower-visitor is urgently needed, and research on the ecology of individual species, on communities of flower-visitors and on interactions between species should be encouraged and promoted.

The usual assumption is. that honeybees interact with native flower-visiting fauna by competing for floral resources and that, this leads to a reduction in the numbers of native animals living in an area. This suggests that food resources rather than some other factor (predation, parasites, weather, nesting habitats) limits the survival and reproductive outputs of these animals. Ideally the importance of each of these factors needs to be assessed as far as developing management programs for particular fauna are concerned. However, a more strategic and focused approach is required to assess impacts of honeybees. Impacts of honeybees ultimately should be measured in terms of changes in population sizes of native fauna in response to changes in numbers of honeybees. However, measuring the population sizes of floral visitors living in an area is fraught with difficulty, particularly for highly mobile animals like floral visitors. Consequently accurate estimates of the numbers of floral visitors living in an area are often impossible to gather. Furthermore many invertebrates are small, cryptic and often difficult to see and identify, even when foraging. Some other parameters are therefore needed for assessing potential impacts.

If floral resources are limiting at a particular plant species and honeybees start to exploit those resources more extensively, then that should reduce the quantities of food being harvested from that plant by native animals, and should change the numbers of native fauna using that plant and/or change their foraging behaviour. Thus the first step in assessing impacts of honeybees on native flower-visiting fauna should be to determine if the quantities of floral resources that native flower-visitors are harvesting from plants changes with changes in honeybee numbers at flowers. If consumption of floral resources by native fauna declines with increases in honeybee numbers then that indicates a competitive interaction. Such calculations also permit some estimate of the extent to which the population sizes of native fauna might be reduced. For example, if the share of resources for one species changes from 30% to 15% then population sizes of this species should decline by 50%. Within a community of many interacting species, the share of resources may actually increase for some populations and decrease for others in the presence of honeybees but the overall share of resources for native fauna declines.

Estimating the quantities of nectar and pollen being consumed by different fauna involves measuring the nectar and pollen contents of flowers at different times of the day, determining the frequency of flower visits by different taxa throughout the day (and any patterns with respect to use of different floral stages), and measuring the quantities of nectar and/or pollen that each removes during a visit to a flower. More detailed descriptions of methods are provided in Paton (1982a, b, 1985, 1986, 1990, 1991, 1993).

The second stage of the assessment process should document the effect of food losses on population sizes, reproductive performance and/or behaviour of native fauna. Some of this is covered by the methods used for recording changes in resource consumption. For example, to estimate food consumption, the numbers of native fauna foraging at flowers, and the speed and efficiency with which they handle flowers are recorded at regular intervals throughout the day. Thus changes in the numbers of native fauna foraging at flowers (and/or the frequency with which they visit flowers) with changes in honeybee numbers are already being recorded. However, to fully assess and understand the mechanism(s) of any competitive interaction further information is often required.

Native fauna could respond in a variety of ways to food losses. They could:

  1. shift to other areas that are not being exploited as heavily by honeybees;
  2. switch to using other plant species not being used as heavily by honeybees;
  3. increase the time they spend foraging to partially compensate for reduced food availability;
  4. alter the time of the day when they feed;
  5. consume less food and loose condition;
  6. reduce their requirements for food (stop feeding, aestivate);
  7. produce fewer offspring, smaller offspring and/or cease to reproduce; or
  8. die.

Some of these additional parameters should - also be measured but what is measured will depend on: the species being studied; the status of the population at the time of the study (eg some populations will not be reproducing at the time of the experiment in which case reproductive parameters cannot be measured); and the ease with which these parameters can be scored for a particular species. In some cases detailed observations on marked individuals will be required and this can be difficult for some species. For example, most Australian native bees are small, fly rapidly, are difficult to mark and track, and are easily disturbed when foraging. Added to these technical problems will be considerable variability between individuals which can only be countered by relatively large sample sizes. Similar types of problems will exist for other taxa.

One of the main reasons for these additional assessments is to allow the severity of any interaction to be better understood. For example, if the numbers of a particular taxon decline at a plant following increases in honeybee numbers then that decline could be from mortality or emigration, or simply from native fauna switching to other resources. Shifting to other resources or to other areas is a less extreme response than mortality.

Establishing if all or only some individuals within a population of native animals are affected is also important. For example, competitive interactions with honeybees could result in some individuals within a population being excluded from an area while others remain and experience no reduction in their access to resources. Those that remain may increase their foraging activity such that counts of native fauna at flowers before and after an increase in honeybee activity remain similar, suggesting no change in numbers of native fauna. If the impact of honeybees was then measured in terms of reproductive rates (number of offspring produced per adult remaining) or as body condition (body mass) then little difference between control and experimental sites in these parameters might exist as well, since the remaining adults still have the same quantity of resources available to them. The conclusion drawn from such data would be that there was no significant effect of honeybees on this population, yet the numbers of adults remaining at the experimental site(s) had actually declined as had their total share of resources. If, however, the individual animals were marked and records kept of the numbers of marked animals resighted before and after an increase in honeybee activity, this reduction in numbers should be detected (ie significantly fewer marked animals would be resighted at experimental sites compared to control sites). Recording foraging activities of some of these marked individuals would also detect increases in their foraging activity at the experimental sites and indicate further the significant effect of honeybees. Thus conclusions reached from studying interactions between honeybees and native fauna may differ depending on which parameters are measured.

This hypothetical case illustrates the difficulty in measuring impacts of honeybees on native fauna and the need to measure a suite of parameters if valid conclusions are to be drawn. Studies should first establish that consumption of floral resources by honeybees reduces the quantities of food available for native fauna and then establish how this loss of food influences various components of the foraging behaviour, condition, abundance and/or reproductive performance of native fauna.

Experimental manipulations

When ecologists test for an effect of honeybees on the abundance, behaviour and reproductive performance of native biota, they usually introduce a number of hives to one or more experimental areas and compare responses of plants and animals near these introductions to those of plants and animals distant from them (control areas). In many cases there is already a background level of honeybees present and the introduction of further hives of honeybees may not result in any significant increase in the numbers of honeybees foraging on the target plant in the experimental plots (eg Schaffer et at. 1983). Honeybees from individual hives often forage widely and out to distances of at least 2 km from their hives. This is equivalent to a minimum area of at least 12 km². Furthermore these bees may not be evenly distributed over this area (eg Visscher and Seeley 1982). Conceivably then, the introduction of ten hives to an experimental area may not increase the effective hive density in a region by more than 0.8 col/km² (0.008 col/ha) and not result in a substantial increase in resource consumption by honeybees in the experimental plots. If control areas were nearby (even within 5 km) then these too could conceivably experience increases in honeybee activity similar to the experimental areas.

An added problem with these experimental manipulations is that simply increasing the numbers of honeybees working flowers may not lead to any significant increase in the quantity of food harvested by honeybees, particularly if honeybees are already removing a large share of the resources. For example, relationships between the consumption of floral resources and numbers of honeybees working flowers may not be linear (eg Figure 2). When honeybee densities are low small increases in the numbers of honeybees working flowers may lead to substantial increases in the share of resources that they harvest. However, when numbers are high further increases in the numbers of honeybees working flowers may only result in negligible increases in the quantities of resources that honeybees consume. Thus, simply adding hives to an area may not cause any significant increase in competitive interactions. Before drawing any conclusions from manipulative experiments, ecologists need to show that those manipulations have actually worked and influenced the share of resources being secured by native fauna. So far, few studies have done this.

Given that future management of honeybees is either going to maintain the current levels of honeybees in conservation areas or attempt to reduce them, experimental manipulations to measure effects of honeybees on native flora and fauna should probably consider removing honeybees from an area rather than adding more hives. Although managed hives can (in theory) be excluded from sensitive areas and from surrounding buffer zones, techniques for removing feral colonies of honeybees from such a wide area are poorly developed. Thus, reducing the numbers of honeybees working flowers over a wide area is not possible at present. However, honeybee numbers can be reduced over smaller areas simply by trapping and removing honeybees that visit and attempt to forage at flowers in experimental plots while not doing this for control plots. In most cases after an initial period of removing honeybees, the numbers of honeybees arriving at these experimental plots declines. This initial decline happens because individual honeybees usually return to forage in the same patch of flowers and once these regular visitors have been removed the numbers of honeybees arriving at plots is greatly reduced (Paton pers. obs.). New honeybees, however, continually recruit to patches of flowers but usually in small numbers, so some vigilance is required to maintain an area relatively free of honeybees.

Figure 2
Figure 2. Models of the quantities of nectar and pollen removed by honeybees from Eucalyptus cosmophylla (a) and Correa reflexa (b) respectively as a function of the numbers of honeybees and honeyeaters visiting flowers. In both models the amounts of resource consumed by honeybees are plotted for three levels of bird visitation (5, 10 and 30 visits/flower/day for E. cosmophylla and 0.5 1.5 and 2.5 visits/flower/day for C. reflexa).In these models the proportions of floral resources not taken by honeybees were taken by birds. The models are based on observed patterns of visitation, measured efficiencies of nectar or pollen removal from flowers for honeybees and birds, and take into account diurnal patterns for nectar secretion or anther dehiscence. Details of these measurements and assumptions used for these models are given in Paton (1990).
Spatial and temporal scales

The other major problem in assessing impacts of honeybees on native flora and fauna is the length of time that native fauna require to respond to changes in honeybee densities and the size and distance between replicate plots. These spatial and temporal scales are always difficult to define, are likely to differ from taxon to taxon and differ with the type of experiments being attempted (removal versus addition of honeybees). For example, most solitary native bees probably forage within a few hundred metres of their nests (eg Donovan 1980), although hard evidence for this is lacking. If so their performance in an area is influenced by changes in resources at that scale. However, the numbers present in one year may be determined by conditions in the previous year since those conditions will determine the numbers of eggs that were laid, the quantities of food provided for larvae and the numbers of adults that subsequently emerge. Thus any impacts of honeybees on native bees in one year may not be expressed until the next generation -of adults emerges. The foraging range of honeybees adds a further dimension to the problems of spatial scales. Although individual honeybees may forage in relatively well-defined and small areas that are possibly smaller than the areas used by individual native bees, honeybees from a single hive collectively forage over a wide area. As a result honeybees are likely to be influenced by the availability of resources over an area of at least 10 km² and potentially up to 300 km² when conditions are poor. Independent replicates therefore often need to be at least 2 km, if not further; apart. Spatial scales may be even greater for mobile species like honeyeaters which may move on a regular basis over hundreds or even thousands of square kilometres. Their abundance and behaviour in an area may be influenced by the availability of resources in that area as well as by the availability of complementary resources in distant areas at the same or other times of the year. Executing manipulative experiments and measuring responses of native fauna at these larger landscape scales, however, will be difficult and a commonsense approach to setting spatial and temporal scales for experiments is required. For plants the appropriate minimum spatial scale is probably determined by the neighbourhood sizes of the plants. Spatial scales for studies on native fauna are more difficult to define and so a range of spatial scales should probably be considered. Minimum sizes of experimental plots should probably exceed the area that an individual usually uses when harvesting its food requirements at the time of the experimental manipulations.

As far as temporal scales are concerned, studies should be conducted over several years to account for year to year variations in production of floral resources by different plants, weather conditions and other potentially limiting factors (predators, parasites, pathogens).

Use of hollows by native fauna and honeybees

Potential competition between honeybees and native fauna for hollows has largely been assessed by recording the proportion of hollows that are occupied by different fauna including feral colonies of honeybees at one particular time. The -usual conclusion from these studies is that only a small proportion of the available hollows are occupied and therefore that honeybees were not displacing other hollow nesting fauna from an area (eg Oldroyd et al. 1994).However, most of these studies fail to provide adequate information on the internal characteristics of the hollows being used by different animals, on patterns of use of hollows through time, and/or any relationships between attributes of different hollows and the performance of native fauna to confidently eliminate competitive interactions (the work of Saunders (1979) and Saunders et al. (1982) being exceptions). A more detailed and rigorous approach is now required to properly document the availability and use of hollows by honeybees and native fauna in a region. These also need to be supported with experimental studies that test for any negative effect of honeybees on hollow-frequenting fauna.

Three types of descriptive studies are warranted:

  1. studies on the population ecology of the hollows themselves;
  2. studies on the patterns and dynamics of hollow occupancy in a region; and
  3. studies on the ecology of specific hollow frequenting fauna.

Each type of study has a different focus. The first focuses on hollows and is concerned with providing information pertinent to the management of hollows in an area. Amongst the questions that need to be answered are: How many hollows are present in an area? How do they differ in size and shape? How do the numbers, sizes and shapes of hollows change through time? At what rate are new hollows being formed and hollows of different sizes and shapes being lost (eg branch fall, tree fall, tree removal, natural decay)?

Studies on the patterns and dynamics of hollow-occupancy are aimed at documenting the frequency with which hollows are used and any successional or temporal patterns in use. For example, hollows that have been used by honeybees may no longer be attractive to other wildlife, even after the honeybees have died or been removed. Native fauna may also only use particular types of hollows and not others, and they may or may not show strong fidelity by breeding in the same hollow in consecutive years.

These descriptive studies on hollows and hollow-occupancy should consist of

  1. thoroughly examining trees for entrances to cavities (not just making visual assessments from the ground);
  2. examining, describing and measuring the internal features of each cavity (using a battery-powered arthroscope or similar device) and so truly assessing their suitability for use by different fauna;
  3. recording through time the use of cavities by different animals (ie either inspecting hollows on a monthly basis (including noting presence of any faeces and nest material) or using remote sensing techniques at entrances to record use, or observing hollow trees for animal activity (eg Smith et al. 1989);
  4. re-measuring hollows on a regular basis (ca every 3-5 years) to determine how quickly hollows change in size and shape;
  5. re-examining trees for the presence of hollows every 3-5 years to determine the rate of production and attrition of hollows within an area through time; and
  6. ascertaining if younger trees are still being hollowed out by termites or attacked by fungi so determining, in the long-term, the rate at which new hollows may be produced in an area (eg Mackowski 1984).

Collecting these details may be difficult if the trees are tall or in inaccessible terrain and suitable study sites may need to be selected' on logistical grounds (ie ease of access to trees). Cherry-pickers and similar mechanical equipment may be needed to gain access to the upper branches of trees to facilitate inspections of hollows, and coress made into the heartwood to assess presence of termites or fungi.

Finally, studies on the ecology of specific hollow-frequenting species should assess whether the availability of hollows or some other factor limits the population sizes of these species. In this case the focus of any research is on the particular species and not on hollows per se.However, the work should include identifying any patterns between the shape, size, location and orientation of hollows and the ability of occupants to survive and reproduce in those hollows (eg as Saunders (1979) has done for White-tailed Black Cockatoos). Various species of rare or threatened cockatoos and parrots are obvious choices for these types of studies. For these birds, assessments should include identifying the factors that cause any nest failures (predation, starvation, disease, weather, desertion) as well as recording the frequency of any agonistic interactions between individuals at or near hollows. If hollows are limiting then intraspecific and interspecific interactions at hollows should occur frequently, particularly early in the breeding season. Furthermore, if hollows are limiting then some individuals capable of reproducing should not breed. Note that the presence of non-breeding birds in a population does not necessarily mean that hollows are limiting, since other factors may have prevented those birds from breeding but if the whole population is breeding then availability of hollows is unlikely to be a limiting factor. Much of this work involves detailed observations on individual birds to establish if they are not breeding and to assess the quality and proximity of the food supply being used during breeding. If individual birds spend large amounts of time foraging then that is indicative of a poor quality food supply and suggests that food rather than hollows may be limiting.

The above descriptive studies are time consuming, tedious and expensive to implement, yet these details are needed to properly assess impacts of honeybees on hollow-nesting fauna and to properly manage hollows and native wildlife in the future. Ultimately the impact of feral honeybees on hollow-nesting fauna will need to be tested experimentally but detailed information on hollow-use will still be needed to assess the outcome of these manipulations.

Two experimental manipulations are possible. The first manipulation involves removing feral colonies from some plots of woodland and not others and then recording the responses of native fauna to this manipulation. In this case the densities of native fauna using hollows in experimental plots are measured before and after the removal of feral colonies and compared with similar data collected on control plots where feral honeybees have not been removed. If honeybees have a significant effect on native hollow-nesting fauna then there should be an increase in the numbers of native fauna using hollows in those plots where feral colonies have been removed. The second manipulation involves adding nest boxes of appropriate dimensions to some plots and not others and again comparing the responses of native fauna at these plots with data for control plots -where no nest boxes have been added. In both cases detailed information on the numbers of hollows being occupied by different fauna is required both before and after the manipulation. Simply recording whether native fauna use nest boxes or use the hollows that have been vacated by honeybees is not sufficient to test for competitive interactions. Native fauna may simply use these `new' hollows by chance instead of using others in the plot, and so there may be no change in the overall densities of these animals following the manipulation. A change in overall nesting densities or total numbers of hollows that are occupied is the key variable.

The temporal and spatial scales for these experiments may provide a further challenge. Individual plots may need to be large (5 ha or more in area) to provide sufficient numbers of native fauna to be able to measure their responses with some degree of statistical power, and the manipulations may need to be maintained for several years to provide sufficient time for populations of native animals to respond.

Population dynamics of feral honeybees

Studies on the dynamics of feral populations of honeybees are required to:

  1. establish the factors (eg food, water, weather, hollows etc.) that limit the numbers of colonies and their sizes in different areas;
  2. measure rates of turnover of colonies in an area (including patterns associated with different-sized hollows); and
  3. record seasonal patterns in resource use, colony strength and survival.

Such information will be useful for implementing programs to control feral colonies of honeybees. For example, feral colonies of honeybees may be more susceptible to eradication programs when food is not abundant.

This research involves:

  1. regularly searching an area for feral colonies to identify any new colonies that have established since the last search;
  2. recording whether previously located colonies were still alive;
  3. collecting information on the types of hollows being used by feral colonies;
  4. identifying any patterns to the survival of colonies occupying different hollows;
  5. identifying any patterns to the recruitment of new colonies (eg whether hollows that were previously used by honeybees were more likely to be reoccupied by honeybees than those that had not been used);
  6. recording the activity of colonies (ie numbers of foraging honeybees returning per minute) at different times in the year;
  7. collecting nectar and pollen samples from honeybees returning after foraging at different times of the year to determine the primary floral resources being used to support feral colonies; and
  8. estimating the quantities of nectar and pollen being produced by the major plants at different times of the year and relating these to the foraging activities of feral colonies and the colonies gain or loss in strength since the last assessment.

Methods for collecting these data have been described in earlier sections of this report.

Research on the efficacy of methods to control honeybees

At present the usual method for controlling feral colonies of honeybees is to individually poison each feral colony that is found. Shelltox pest strips and other over-the-counter insecticides have all been used to some extent to kill individual colonies. This is labour intensive and not an effective method of controlling or eradicating feral colonies of honeybees over a wide area. Some efficient broadacre technique needs to be developed to help land managers remove feral colonies from sensitive areas. However, considerable research will be required before any broadacre methods can be introduced. The most likely method(s) of control will involve using baits laced with certain chemicals that honeybees take back to the hive that eventually kill the hive. There are three areas requiring research:

  1. the impact of the baiting program on nontarget animals both from primary and secondary poisoning;
  2. the success of the baiting program in eliminating feral colonies from a region (ie what proportion of colonies within a specified distance from a baiting station are destroyed); and
  3. the speed with which the baited area is recolonised by feral honeybees.

Initial research should place out attractants for foraging honeybees (eg sugar syrup, water, bran) and record the frequency with which honeybees and native fauna attend these feeding stations. The location of dispensers (eg hanging from branches, in the shade, out in the open, close to flowering plants), the type of dispenser and the concentration of the food should all be varied to determine the conditions that are most attractive to honeybees and least attractive to native fauna. This should also include documenting seasonal patterns of attendance at feeding trays. Danka et al. (1992) and Scriven (1995) both report that when floral resources are scarce large numbers of honeybees attend feeders, and when large numbers are attending feeders other animals are excluded.

Once the best methods and times of the year to attract honeybees but not non-target animals have been identified, the next stage in the assessment process should involve assessing various poisonous chemicals for use in killing feral colonies. A range of poisons could be considered but probably acephate, dimethoate and dichlorvos or similar chemicals should be tested first (eg Waller and Barker 1981; Williams et al. 1988, 1989; Woodward and Kassebaum 1991; Danka et al. 1992).There are many factors that need to be considered, including the dosage used with the attractant, the quantities needed to kill feral colonies and the subsequent fate of the poison in the environment. First, the dosage cannot be too strong or otherwise foraging honeybees are killed before they have had time to return to the hive. Second, if the dosage is too low, the amount of poison arriving back at a hive may never be sufficient to kill that colony. Some initial information can probably be obtained by allowing individual honeybees to consume different doses of particular poisons inside enclosed chambers and then recording the length of time that they survive (eg Waller and Barker 1981).Dosages that allow foraging honeybees to survive long enough to return to their hives should then be used.

Having established the optimum concentration for the poison in the baiting solution, the next stage of the assessment process should involve examining the quantities needed to kill feral colonies. Initially some trials could be performed on managed colonies of honeybees by placing quantities of the bait solution in a small but open container inside the hive-where only honeybees from that hive have access to the poisoned bait. By measuring the quantities of the poisoned bait left in the container and the quantities of poison remaining in frames once the colony has died, some measure of the quantity of poison needed to kill hives should be obtained. Some trials on feral colonies may also be possible by placing a container with poisoned bait at the entrance or near to a feral colony and recording the amounts that had been consumed before the colony died.

Considerable research on developing chemical methods of controlling feral honeybee colonies (particularly Africanized honeybees) is currently being conducted in North America and some of this research will provide guidelines on dosages and quantities for some chemicals (Williams et al. 1988, 1989; Loper and Sugden 1990, 1994; Danka et al. 1992).Danka et al. (1992), working in Texas and Louisiana, have recently estimated that at least 25 mg of acephate was needed to guarantee the mortality of a colony of honeybees. They also showed that this quantity could be successfully delivered if at least 100 honeybees from a colony were foraging at a baiting station when a 50% sucrose plus 10% honey solution was replaced with one that was laced with 500 mg L-1 acephate for 20-30 minutes.

Depending on the chemical that has been used, there may be a risk of secondary poisoning if native animals (eg ants) scavenge dead bees, honey, wax and/or pollen from a poisoned colony (Danka et al. 1992; Scriven 1995). Two assessments are required. First, the animals that are likely to feed on the remains of a feral colony need to be identified. Second the fate, concentration and stability of the poison that arrives back at a feral colony needs to be measured to determine the risk of secondary poisoning for non-target animals. Some information on the animals likely to scavenge at honeybee colonies could be obtained by recording the species that attend live colonies and colonies that have died of natural causes or have been killed by other means: Ideally the chemicals used to poison honeybees should breakdown rapidly to nontoxic substances. Samples of dead bees, honey, wax and pollen could be taken at regular intervals after the colony has been killed and assayed for any residual toxicity. Danka et al. (1991,1992) have measured residues - of acephate and methamidophos from dead bees and from the honey-wax matrix of managed colonies killed with acephate and suggested that residue levels and decay trends varied greatly among colonies. Having selected an appropriate poison and method for dispensing the poison, the final stage in the assessment process involves field trials to test the efficiency of the method in eradicating feral colonies over a broader area. This should involve:

  1. determining the numbers of feral colonies present in a number of areas (see above) at various distances away from a central location;
  2. attracting honeybees to a feeding table stationed at that central site;
  3. having attracted honeybees to the feeding station, the poison is added to the food being provided and the bees allowed to continue to forage; and
  4. the survival of feral colonies present in the various plots at different distances from the baiting station is subsequently scored.

Based on these results some indication of the numbers of baiting stations needed per square kilometre can possibly be calculated to provide a reasonable cover within a selected area. For example, Danka et al.(1992) found that most colonies within 200 m of baiting stations in Louisiana and Texas usually attended feeders in sufficient numbers (>100 individuals/ colony) to be killed when exposed to acephate. Attendance by honeybees from colonies beyond 200 m, however, was usually not sufficient to deliver fatal amounts of acephate. These data suggest that about 10 baiting stations per square kilometre would be required at these American sites to provide reasonable coverage.

The speed with which baited areas are recolonised should also be measured to assess how frequently control programs need to be implemented. Placing decoy hives with attractant lures (eg Schmidt and Thoenes 1992; Winston and Slessor 1993) in areas during the swarming season may help to restrict the rate at which feral colonies recolonise areas, as well as allowing swarms to be easily removed.

Assessment of resource use by commercially-managed apiaries

Amateur and commercial beekeepers maintain over 500,000 hives of honeybees in Australia. Commercial apiarists and some amateur apiarists shift large numbers of hives into areas to exploit 2-4 month long peaks in the flowering of key native plants. These key plants are often species of Eucalyptus but can also include a variety of other shrubs and small trees (eg Eucryphia, Banksia, Dryandra) that because of their prominence can produce large quantities of nectar and/or pollen on a per unit area basis. Beekeepers often argue that their bees simply exploit the surplus floral resources being produced. This may be the case but this use of floral resources needs to be properly assessed. First, large numbers of hives are introduced at specific locations that may result in densities of honeybees that are much higher than background densities of feral honeybees at least on a local scale and, as a result, they may have comparable impacts on resources and native taxa. Second, commercially managed honeybees will visit not only the flowers of abundant plants, but also less abundant plants where surpluses may not exist and where specific native fauna may be severely affected. Third, there are at least qualitative differences in the magnitude of the floral peaks from one year to the next (at least as measured in terms of honey production) in many areas (eg Berkin 1987, Manning 1992) and perhaps in the timing of flowering such that the shifts of bees into and out of areas may not always match the periods of peak flowering. Furthermore, there is an assumption that there is a staggered sequence of peaks in floral resources that allows beekeepers to shift from one resource or area to another without experiencing any lean periods. Such a staggered and complete annual sequence of peak resources may not exist. Turner et al.(1972) in fact recommend that hives should not be shifted from one honey flow to the next but rested between flows to allow them to rebuild hive strength. Fourth, the importance of floral peaks both in supporting native fauna and in the reproductive biology of the plants is poorly documented let alone understood.

The following items need to be measured to address some of these deficiencies and to allow sensible management plans to be developed.

  1. How closely do commercial beekeepers track peaks in floral resources?
  2. Which plant species provide the bulk of the resources used by commercial apiaries?
  3. How does the flowering of these plants (timing and intensity) vary from year to year?
  4. What proportion of the floral resources produced by these plants are harvested by honeybees?
  5. What other plant species are used during these peaks and what proportion of their floral resources are harvested by honeybees?
  6. Is there an annual succession of flowering peaks in different areas that can be exploited by commercial beekeepers?
  7. How do native flower-visiting fauna respond to these influxes?
  8. How does seed production vary across the flowering season and with influxes of commercially-managed honeybees?

These initial aims simply describe the use of resources and do not aim to assess whether the commercial industry has a negative effect on the conservation values of natural areas or resources being used. This basic information is critical for selecting those - aspects where impacts are most likely (ie particular sites, times of year, primary versus secondary plant species and/or particular fauna).

Two approaches should be taken to collecting these baseline data:

  1. studies that track and document the movements and resource use of a number of commercial apiaries (ists); and
  2. studies that track resource production in reserved areas where commercial loads of honeybees are placed at certain times of the year.

These studies should span at least three if not five years to document annual variation - in resource production and patterns of use. The most efficient strategy would probably involve selecting a number of apiarists who have kept accurate records of their movements and performance. From these records, 4-10 sites that are used regularly by them should be selected for study. At these sites, the quantities of floral resources being produced would be measured at regular intervals throughout the year to document the seasonal patterns in food availability. This would require determining the abundance of plants at various distances from the site where hives are placed using quadrats and random stratified sampling. Having established the density-of the plants - future work should only involve counting the flowers present on - subsamples of plants, measuring nectar and pollen production at samples of flowers, diurnal patterns to availability and the quantities being removed by honeybees and native fauna before, during and after the commercial apiaries have been placed in each of those areas. Simple techniques for assessing resource use have been outlined above. Note, however, that - because the density of honeybees may decline with distance from an apiary some measure of resource use as a function of distance from an apiary should also be attempted as part of these studies.

Three criteria could be used to assess the likelihood that influxes of commercial loads of honeybees affect native flora and fauna:

  1. the quantity of resources consumed by native taxa declines when commercial apiaries are introduced;
  2. the change in numbers and or behaviour of native pollinators/flower-visitors with the introduction and subsequent removal of commercial loads of honeybees; and
  3. the change in production of seeds by the plants in response to the arrival and departure of commercial loads of honeybees.

Ideally these responses would need tobe compared with similar data collected at equivalent control areas where no commercial loads of honeybees had been placed to properly test for an impact of commercially-managed honeybees on natural systems. However, even without the control areas some useful information on the likely affects of beekeepers' honeybees on natural systems would be obtained. Such information would clearly help in developing appropriate management programs to reduce potential impacts.

Integration of research projects and priorities

The research programs that have been outlined above are all beneficial to the long term management of biotic resources in Australia. However, with the current economic climate funding all of these programs is unlikely. So which research programs should be given priority and where should those programs be carried out?

Most land managers indicate that they require information on:

  1. the impacts of honeybees on native flora and fauna;
  2. the distribution, abundance and population dynamics of feral colonies of honeybees; and
  3. methods of efficiently and effectively removing feral colonies from an area, before implementing management programs.

In most cases the areas that they would target would be those areas with high densities of feral colonies and/or those areas where there was some concern that honeybees might impact on certain rare or endangered wildlife, mainly hollow-nesting parrots and cockatoos (Regent Parrot in Victoria; Black-Cockatoos in South Australia). The only other situations where feral colonies are given some management attention are areas where feral honeybees cause problems for humans.

The numbers of feral colonies present in an area, however, is not necessarily the best criterion for selecting sites in which to do initial work. Flora and fauna in areas with low densities of feral colonies may be just as heavily affected by honeybees, if not more so, than flora and fauna living in areas with high densities. Also, available evidence suggests that most hollow-nesting fauna are not likely to be adversely affected by honeybees (eg Rowley 1990; Oldroyd et al. 1994) but the impacts of honeybees on native flora and flower-visiting fauna could be severe based on the frequency with which honeybees visit the flowers of a wide range of plants (eg table 6). On these grounds research should concentrate on documenting the impacts of honeybees on native flora and flower-visiting fauna rather than on hollow-nesting fauna. Note that where there is concern that honeybees may be impacting hollow-nesting fauna some management actions are being implemented as part of the management programs for those endangered wildlife (eg Glossy Black Cockatoo on Kangaroo Island) and so the concern is alleviated to some extent.

Opportunities to integrate several research programs in the one area should not be ignored, since this will increase cost efficiency. For example, research programs that develop techniques for baiting feral colonies must assess the efficacy of baiting programs on feral populations. This requires information to be collected on the densities of feral colonies before and after the introduction of a baiting program. Thus as part of this program some information could be collected on the types of hollows being used by feral colonies and the proportion of hollows in an area that they occupy. Studies on population dynamics of feral colonies and rates of recolonisation could be incorporated into programs measuring the efficacy of baiting programs. Furthermore, studies on the impacts of honeybees on native flora and flower-visiting fauna could be conducted in the same areas as baiting trials providing a further opportunity to experimentally test the impacts of honeybees on these taxa. These studies could also be conducted in areas where there are concerns about particular hollow-nesting fauna providing the potential for a further benefit.

By concentrating and integrating research programs in a few areas the benefits of any experimental manipulations will be maximised. Which areas should be chosen is difficult to determine. There may be some benefit to working in areas with high densities of feral colonies simply because of advantages with larger sample sizes (of feral colonies) and there may also be some merit in building on the databases that have already been collected in some areas (eg Oldroyd's work at Wyperfeld; Paton's work at Flinders Chase on Kangaroo Island; or around the nesting sites of Glossy Black Cockatoos on Kangaroo Island) rather than starting entirely from scratch in new areas. The choice of areas used for study may ultimately-depend on the availability and interests of research staff willing to be involved in this type of research.

In order of priority, the areas for research each should be:

  1. to promote studies that assess the impact of honeybees on native flora and flower visiting fauna to provide a firm basis for implementing and justifying any programs of control; and
  2. to develop cost effective, environmentally safe methods of eradicating feral colonies of honeybees from selected areas.

These two research areas should be given precedence but less intense studies on the population dynamics of feral colonies and on patterns of hollow use by honeybees and native fauna should be incorporated into these programs if at all possible.

Possible sources of funding for these programs include the Australian Nature Conservation Agency (Invasive Species Program, States Cooperative Assistance Program), World Wide Fund for Nature, Australian Heritage Commission (National Estate Program), various state government grants, the Australian Research Council, and the Honeybee Research and Development Council. Of these, probably only the Australian Nature Conservation Agency and Australian Research Council have the resources to provide sufficient financial support to fund these research programs adequately. The Honeybee Research and Development Council is more likely to fund projects documenting the migratory patterns and use of floral resources by commerciallymanaged honeybees.