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

3. Interactions Between Honeybees and Australian Biota

The most important aspect of an assessment of honeybees in Australia is their interactions with the Australian flora and fauna. Some information on these interactions and some evidence for competition between honeybees and selected native fauna are available.

Which plants and animals interact with honeybees over floral resources?

Information on the plant species used by honeybees as sources of nectar, pollen or resin are limited to state or regional lists of plant species of economic importance to commercial beekeepers and to a few more detailed community studies. Most states have publications that list those plant species that produce substantial quantities of nectar or pollen that can be profitably exploited by commercial beekeepers (Purdie 1968; Goodman 1973; Clemson 1985; Manning 1992). Species of Eucalyptus, Melaleuca, and Banksia are usually listed if they are abundant within a particular region and produce reasonable quantities of nectar and/or pollen on a more or less regular basis. Some indication of annual variation in yield is often given. Less prominent plants that provide few resources for honeybees in these regions are not listed.

More detailed community studies reveal that honeybees interact with a large proportion of the plant species present in temperate heathlands and woodlands. In southern parts of South Australia, honeybees have been recorded visiting over 180 native plant species, approximately half of all the plant species examined during their flowering seasons (Paton 1993, unpubl.). These plants came from 34 families and 86 genera. They ranged from small herbs to large trees and included plants that were pollinated by wind, insects, birds and mammals. Bell (1987), van der Moezel et al.(1987), Wills (1989) and Wills et al.(1990) recorded honeybees foraging at 136 native species from 67 genera and 30 families in kwongan sandplain areas of WA. This was over 30% of the 413 plant species present in the area, though not all of these species were observed when in flower. Again, wind-, insect- and bird-pollinated plants were visited by honeybees. The methods used in these studies favoured detection of honeybee visits to species that were widespread and/or had substantial floral displays. Some of the less abundant species may not have been surveyed adequately to establish if honeybees visited their flowers.

Information on the use of Australian plants by honeybees in other communities is generally lacking, though G Williams (Aust Museum) holds some unpublished information on a few plants used by honeybees in subtropical coastal rainforests. In other countries, honeybees have been recorded visiting similar numbers of plant species. For example, Villanueva (1984) reported that European honeybees visited 185 plant species in lowland forest and agricultural areas of Veracruz, Mexico and Roubik (1988, 1991) estimated that honeybees harvested pollen from at least 142 plant species at sites in Panama and were probably visiting 25-30% of the flora.

Table 5 lists over 200 Australian plant genera known to be visited by honeybees in Australia. The table is not based on an exhaustive survey but clearly shows that diverse taxa are directly involved in interactions with honeybees. Other plant species may interact indirectly with honeybees if their native pollinators also use plants exploited by honeybees and the behaviour and abundance of native pollinators are altered as a result. No studies have considered those types of interactions in an Australian context (but see Hopper 1987).

There is an equally wide diversity of native fauna that use the same plant species for floral resources as honeybees (Armstrong 1979; Anderson 1989; Wills et al.1990; Pyke 1990; Ettershank and Ettershank 1993; Paton 1993). Knowledge of vertebrate plant 21 use far exceeds knowledge for invertebrate use. Conservative estimates suggest that thousands of native invertebrates (beetles, butterflies, moths, bees, flies, wasps, ants) now share food resources with honeybees at flowers, as do over 100 species of vertebrates. In many cases specific identities of the insects involved in interactions are not known except in a scattering of studies where insects have been collected and identified (Hawkeswood 1981a,b; Bernhardt et a1.1984; Ireland and Griffin 1984; Kenrick et al.1987; Wills et at. 1990; Ettershank and Ettershank 1993; O'Brien and Calder 1993; Paton and Jansen unpubl., G Williams unpubl.).

Although not well documented many of the species of animals involved in these interactions harvest floral resources from a variety of different plant species (Paton and Ford 1977; Armstrong 1979; Bernhardt and Walker 1984; Bernhardt e t at. 1984).

These qualitative observations indicate that a great diversity of Australian plants and animals interact with honeybees by sharing floral resources. This diversity greatly complicates any assessments of impacts of honeybees. In assessing impacts of - honeybees in an area a representative selection of plant and animal species should be studied. Whether these interactions are harmful will depend on whether resources are limiting and what share of the resources are consumed by honeybees.

Table 5
List of Australian plant genera visited by honeybees. Genera shown in bold provide significant quantities of floral resources to commercial apiaries.
Aizoaceae Carpobrotus, Disphyma
Amaranthaceae Ptilotus
Anacarcdiaceae Euroschinus
Apiaceae Trachymene
Araliaceae Schefflera
Arecaceae Archontophoenix
Avicenniaceae Avicennia
Campanulaceae Wahlenbergia
Casuarinaceae Allocasuarina, Casuarina
Chenopodiaceae Halosarcia, Rhagodia
Chloanthaceae Pityrodia
Cornpositae Brachycome, Calotis, Olearia, Podotheca, Senecio, Sonchus
Convolvulaceae Calystegia
Cruciferae Lepidium
Cunoniaceae Ceratopetalum
Cyperaceae Cyperus, Mesomelaena, Lepidosperma
Dillenaceae Hibbertia
Droseraceae Drosera
Ebenaceae Diospyros
Eleaocarpaceae Eleaocarpus
Epacridaceae Acrotriche, Andersonia, Astroloma, Brachycoma, Epacris, Leucopogon, Styohelia, Woollsia
Escalloniaceae Cuttsia
Eucryphiaceae Eucryphia
Euphorbiaceae Adriana, Bertya, Beyeria, Micrantheum, Phyllanthus, Ricinocarpus
Flacourtiaceae Scolopia
Frankeniaceae Frankenia
Geraniaceae Geranium, Pelargonium
Goodeniaceae Brunonia, Dampiera, Goodenia, Scaevola, Velleia
Gyrostemonaceae Gyrostemon
Haemodoraceae Anigozanthos, Conostylis
Haloragaceae Glischrocaryon, Gonocarpus, Myriophyllum
Hypoxidaceae Hypoxis
Iridaceae Orthrosanthus
Juncaceae Juncus
Labiatae Ajuga, Hemiandra, Prostanthera, Westringia
Lecythidaceae Planchonia
Leguminosae Acacia, Aotus, Bossiaea, Cissus, Clianthus, Crotalaria, Daviesia, Derris, Dillwynia, Eutaxia, Gastrolobium, Glycine, Gomphlobium, Goodia, Hardenbergia, Jacksonia, Kennedia, Mirbelia, Oxylobium, Phylotta, Platylobium, Psoralea, Pultenaea, Senna, Swainsona, Templetonia, Viminaria
Liliaceae Blandfordia, Bulbine, Burchardia, Caesia, Chaemescilla, Dianella, Dichopogon, Lomandra, Tricoryne, Thysanotus, Xanthorrhoea
Loganiaceae Logania
Loranthaceae Amyema, Lysiano, Nuytsia
Lythraceae Lythrum
Malvaceae Alyogyne, Lavatera, Sida
Melastomataceae Melastoma
Meliaceae Melia
Myoporaceae Eremophila, Myoporum
Myrsinaceae Aegicerus
Myrtaceae Acmena, Agonis, Angophora, Baeckea, Beaufortia, Callistemon, Calothamnus, Calytrix, Chamelaucium, Darwinia, Eremaea, Eucalyptus, Eugenia, Hypocalymma, Kunzea, Leptospermum, Melaleuca, Micromyrtus, Rhodomyrtus, Scholtzia, Syncarpia, Syzigium, Thryptomene, Trystaniopsis, Verticordia, Waterhousea, Xanthostemon
Onagraceae Epilobium
Orchidaceae Dendrobium, Diurus, Drakaea, Eriochilus, Prasophyllum
Oxalidaceae Oxalis
Phytolaccaceae Tersonia
Pittosporaceae Billardiera, Bursaria, Cheiranthera, Pittosporum
Polygalaceae Comesperma
Polygonaceae Muehlenbeckia
Primulaceae Samolus
Proteaceae Adenanthos, Banksia, Cardwellia, Conospermum, Dryandra, Grevillea, Hakea, Isopogon, Lambertia, Macadamia, Persoonia, Petrophile, Stenocarpus, Synaphea, Xylomelum
Ranunculaceae Clematis, Ranunculus
Rhamnaceae Alphitonia, Cryptandra, Pomaderris, Spyridium, Ventilago
Rosoceae Rubus
Rutaceae Acradenia, Baronia, Correa, Diplolaena, Eriostemon, Flindersia, Geljera, Phebalium, Zieria
Santalaceae Choretrum, Leptomeria, Santalum
Sapindaceae Alectryon, Atalaya, Dodonaea, Guioa
Scrophulariaceae Euphrasia, Derwentia, Stemodia
Smilaceceae Smilax
Solanaceae Anthocercis, Solanum
Stackhousiaceae Stackhousia
Sterculiaceae Argyrodendron, Brachychiton, Guichenotia, Keraudrenia, Lasiopetalum, Thomasia
Stylidiaceae Stylidium
Surianaceae Stylobasium
Thymelaeaceae Pimelea
Umbelliferae Actinotus, Apium, Hydrocotyle
Violaceae Hybanthos, Viola
Winteraceae Tasmania
Zygophyllaceae Zygophyllum
Sources: Adams and Lawson 1993; T Bartareau pers. comm.; Bell] 987; Bernhardt et al.1984; Clarke and Myerscough 1991; Clemson 1985; Ettershank and Ettershank 1993; Goodman 1973; Gross 1993; Heard 1993; Hopper 1 980b, 1987; Lamont 1985; Manning 1992; Paton and Jansen unpubl.; Purdie 1968; Pyke and Balzer 1985; van der Moezel et al.1987; G.Williams pers. comm.; Wills 1989; Wills et al. 1990.


Prominence of honeybees at flowers of Australian plants

Three types of data provide information on the prominence of honeybees at flowers of Australian plants:

  1. collections of insects from flowering plants;
  2. counts of insects at flowers; and
  3. observations on the frequency of visits by animals to flowers.

Counts and collections of insects from flowers show that honeybees may account for more than half the insects using the flowers of many plant species (table 6). These collections and counts only estimate the numbers of each taxon that are foraging at a particular time and do not consider differences between taxa-in the speed at which they can visit and shift between flowers. Observations on the number of visits to flowers reveal that flowers are often visited several times a day by honeybees and that honeybees often accounted for more than half of all visits to many flowers (table 6). Given this, honeybees could remove a large share of the floral resources from Australian plants and so potentially compete with native flower -visiting fauna.

Table 6
Prominence of honeybees at the flowers of Australian plants. Data provided in the table illustrate that honeybees are often the most prominent visitors to the flowers of a wide range of Australian plants.
  Honeybee visits to flowers    
Plant species % of specimens, counts or visits a visits/flower/day b Other taxa
visiting flower
Grevillea x gaudichaudii
>26.4i (> 0.78)
bd Taylor & Whelan 1988
Pultenaea elliptica
12 (c)
nb Pyke & Balzer 1985
Eucalyptus gummifera
68 (c)
nb,?bd "
Hakea teretifolia
60 (c)
nb "
Leptospermum squarrosum
92 (c)
nb,?bd "
Angophora hispida
66 (c)
nb "
Banksia ericifolia
25-58 (v)
a,bd,m,nb Paton & Turner 1985
Banksia spinuloso
0-66 (v)
bd Vaughton 1992
Grevillea aquifolium
72 (v)
bd Paton 1985
Callistemon rugulosus
95 (v)
bd Paton 1979
Amyema pendulum
62 (v)
bd "
Eucryphio spp
19-48 (s)
  Ettershank &
Eucalyptus cosmophylla
23-66 (v)
bd,nb Ettershank 1993 Paton 1990
Eucalyptus remotes
57 (v)
bd,nb "
Callistemon rugulosus
38-94 (v)
bd "
Adenanthos terminalis
0-98 (v)
bd "
Acacia paradoxes
44-97 (v)
nb Paton et al. MS2
Banksia marginata
67-92 (v)
bd,nb "
Conospermum patens
13-77 (v)
nb,ins "
Dorwinia micropetala
2-40 (v)
nb,ins "
Daviesia genistifolia
77 (v)
nb "
Eucalyptus baxteri
32-74 (v)
bd,nb, ins "
Grevillea parviflora
>83 (v)
nb "
Hakea rostrato
>75 (v)
nb, ins "
Leucopogon parviflorus
52-86 (v)
nb - "
Melaleuca gibbosa
39-53 (v)
nb,ins "
Orthrosanthus multiflorus
>36-1 00 (v)
nb "
Phyllota pleurandroides
22-96 (v)
nb,ins "
Pimelea flava
67-100 (v)
nb,ins "
Prostanthera spinosa
7-81 (v)
nb,ins "
Pultenaea viscidula
86 (v)
nb "
Swainsona lessertifolia
91-100 (v)
nb "
Xanthorrhoea semiplana
84 (v)
nb,ins "
Hibbertia virgata
75-79 (c)
a,bt,m,s Paton unpubl.
Eucalyptus fasciculosa
18 (c)
a,bt,f "
Acacia myrtifolia
86 (c)
2.1 h
nb "
Acacia pycnantha
100 (c)
Dillwynia sericea
50 (c)
nb "
Brachyloma ericoides
bd "
Pultenaea canaliculata
85-96 (c)
nb,ins "
Pultenaea tenuifolia
48-56 (c)
nb,ins "
Pimelea glauca
0-89 (c)
ins "
Pimelea humilis
0-33 (c)
ins "
Calytrix tetragona
0-100 (c)
ins "
Dampiera marifolia
63-100 (c)
ins "
Dampiera rosmarinifolia
82-100 (c)
ins "
Eutaxia microphylla
0-94 (c)
nb,ins "
Eucalyptus odorata
81 (c)
bd,nb, ins "
Dillwynia hispida
11 -100 (c)
nb,ins "
Scaevola aemula
53-100 (c)
ins "
Lasiopetalum baueri
0-94 (c)
ins "
Acacia calamnifolia
0-75 (c)
ins "
Daviesia benthami
13-14 (c)
nb, ins "
Calothamnus quadrifidus
74 (v)
bd Collins et al. 1984a
(a) c = counts; s = specimens; v = visits
(b) i = visits per inflorescence per day; h= visits per flower head per day; all others are visits per flower per day
(c) a= ants; bd = birds; bt = beetles; f = flies; m = moths; nb = native bees; s= syrphid flies; ins = insects

Availability and consumption of floral resources

The beekeeping industry defends its periodic use of natural resources by arguing that beekeepers only exploit surplus resources during flowering •peaks and so have little effect on natural processes. Feral colonies on the other hand are more likely to affect natural processes because they remain in an area throughout the year, including lean periods when more intense competition for floral resources may exist. This argument suggests that in many natural systems there are periods in the year when the amount of nectar being produced is substantially higher than at other times and that native fauna are unable to fully consume these peaks in resource production.

Floral resources in Australian ecosystems have rarely been measured except for the quantities of nectar available to nectarfeeding birds in a range of heathland and woodland sites in southern Australia (Ford 1979; Paton 1979, 1985, 1986; McFarland 1986a,b; Ford and Paton 1982; Pyke 1983, 1985; Newland and Wooller 1986; Collins and Newland 1986). In most of these areas there were substantial seasonal variations in the quantities of nectar being produced. For example in coastal Banksia heathlands near Sydney production varied 10-100 fold throughout the year, averaging 1-10 g sugar/ ha/day during summer and around 100 g sugar/ha/day for 2-3 months over winter when B. ericifolia was flowering (Pyke 1983; Pyke and Recher 1986). Similarly production of nectar in an area of open forest with an understorey of Banksia spp in New England NP varied from effectively 0 g/ha/day during summer to 1125 g sugar/ha/day for 2-4 months during late winter and early spring although from one day to the next the amounts produced in winter varied dramatically (Ford and Pursey 1982; McFarland 1986a,b). In Banksia dominated heaths and swamps in South Australia and Western Australia nectar production also peaked for 2-3 months during winter at 200- 1000 g sugar/ha/day depending on the location (Newland and Wooller 1986; Paton et at. MS3, unpubl.).

In other areas without a dominance of Banksia, like the Royal Botanic Garden's Annexe at Cranbourne, Victoria, nectar production varied from 12-30 g sugar/ha/day throughout the year as different plants came in and out of bloom (Paton 1979, 1986). In woodland areas near Horsham nectar production usually varied from 10-100 g sugar/ha/day but for short periods of time when certain eucalypts flowered productivities approached 375 g sugar/ha/day (Paton 1985). The eucalypts, however, did not flower reliably in each year. Heathy woodlands adjacent to Banksia swamps in WA produced 10-50 g sugar/ha/day and woodland areas adjacent to Banksia heathlands near Sydney produced from 0-225 g sugar/ha/day with peaks in productivity in these habitats not coinciding with those in the adjacent Banksia habitats (Pyke 1985; Newland and Wooller 1986). Production of nectar in Jarrah forests in WA varied seasonally from close to 0 g/ha/day in late summer to around 150 g sugar/ha/day in late spring when B. grandis flowered (Collins and Newland 1986).

In most of these studies attempts were made (with mixed success) to correlate the numbers of honeyeaters and their calculated energy requirements with the quantities of nectar being produced (Paton 1979, 1985; Ford 1983; Ford and Paton 1985; Pyke 1983, 1985, 1988; Pyke and Recher 1986; McFarland 1986a,b; Newland and Wooller 1986). In some areas there was a close match between the quantities needed to support native fauna and the quantities being produced. However, in other areas at certain times in the year, particularly winter, the production of nectar was higher than that needed to support the populations of native nectar-feeding birds in those areas, suggesting that surpluses do exist at least in some areas in winter.

The notion that surpluses sometimes exist is further supported with observations of nectar dripping from flowering inflorescences of some banksias during winter months. Various hypotheses for this phenomenon have been proposed (eg Carpenter 1978). Such surpluses, however, may be caused by a lack of visitation by native pollinators (mainly birds) rather than excessive production by the plants. In most cases the rate of production on a per inflorescence basis is moderate and averages about 0.2-1 g sugar/inflorescence/day depending on the species (Ford and Pursey 1982; Ford and Paton 1982; Paton and Turner 1985; Paton 1986, unpubl.; McFarland 1985; Pyke 1983, 1988). Suggestions that a B. ornata inflorescence can shed over 300 ml of nectar (ca 60 g sugar) in less than a day are, however, inconceivable (Berkin 1987).

These periods of resource surpluses are sometimes used to argue that populations of honeyeaters are not limited by floral resources and that honeybees pose no threat to nectarfeeding birds (Stace 1988; Manning 1993a). However, this selectively ignores other studies that showed the numbers of birds to be closely linked to the production of floral resources and to the proportion of those resources that the birds could harvest (Paton 1979, 1985).

There are of course a range of possible interpretations for the existence of surpluses at certain times in the year. One is that population densities of birds may be limited by the availability of nectar at other times in the year such that their numbers could never reach the carrying capacity of the winter resources. The imbalance could arise from differential clearance of summer habitats relative to winter habitats but could equally be induced by heavier losses of resources to honeybees during the warmer months. Paton (1985) showed that consumption of nectar resources by honeybees was higher during the warmer months of the year than during winter.

In summary, studies have found that surplus floral resources do exist and more in-winter than at other times in the year. However, these studies have been largely restricted to heathlands or low woodlands where understorey plants produce most of the nectar over winter, and to areas supporting reasonably high populations of honeyeaters. As such these studies may not be representative of-other areas. Furthermore, in most cases resources were only assessed over small selected areas (a few hectares) and the results may not extrapolate to estimate nectar production over more extensive areas.

The production of floral resources by eucalypts has largely been neglected in all of these studies, reflecting the difficulty of measuring floral resources produced by tall trees. Paton (1986) reported that nectar production by several species of eucalypts 26 ranged from 0.5 to 25 mg sugar/flower/day and from 2 to 85 g sugar/tree/day with the most productive species being E. leucoxylon.Ironbark (E. sideroxylon) may produce even greater quantities of nectar. For example, Damon Oliver (pers. comm.) estimated that individual flowers produced about 15 mg sugar/day, and that individual trees could produce 200 g sugar/day with ironbark forests (ca 100 trees/ha) producing 20 kg sugar/ha/day. Given this, periods of resource surpluses may also exist when certain eucalypts are flowering but as yet there have been no quantitative measurements.

For comparison some rough estimates of the quantities of floral nectar harvested by commercial apiaries suggest that during a moderate to high honey flow an apiary of 100 hives produces about 6 tonnes of honey during a 3 month period of which about 4 tonnes would be removed by the beekeeper (eg Manning 1993b). This is equivalent to each hive collecting about 60 kg of honey, or 50 kg of sugar (honey is about 85% sugar and 15% water) during a 3 month period or about 500 g sugar/day. Assuming that these bees forage predominantly within 2 km of their hives (a reasonable assumption if occupied apiary sites are on average 4 km apart) then these honeybees would be harvesting about 50 g sugar/ha/day assuming the apiary was completely surrounded by suitable habitat. Although crude, these rates of harvest can be met by the quantities of nectar being produced during peak flowering periods in some habitats.

A second method has been used to assess nectar availability in some of these bird-plant systems. These assessments have involved measuring the quantities of nectar remaining in flowers at various times during the day and comparing these to either the energetic costs of the birds to harvest those resources or to the quantities in bagged flowers (Ford 1979; Paton 1979, 1982a, 1985; Collins et at.1984a,b; McFarland 1986a; Collins and Newland 1986; Armstrong 1991). These studies found that for some plant species there was little change in the quantities of nectar available at flowers throughout the day and that the energetic returns for honeyeaters to harvest those resources remained high throughout the day. Such occasions, however, were more frequent during winter months consistent with earlier findings of winter surpluses in other plants. At other times, notably the warmest months, the energetic returns to honeyeaters for feeding on nectar were at best only marginally profitable for much of the day. Under such conditions, honeyeaters were often territorial around the densest clumps and/or most productive flowers, indicative of limited food resources (Paton 1979, 1985; Ford 1979, 1981).

A third approach to assessing resource availability has been to calculate the proportion of floral resources that are being consumed by different floral visitors at selected plants. Most studies have only considered the quantities of nectar being consumed by different floral visitors at plants largely pollinated by birds. At these plants honeybees consumed 14-97% of the nectar being produced (Paton 1979, 1985, 1990; Bond and Brown 1980; Collins et at.1984a; table 7). In most cases birds began foraging earlier in the day than honeybees. At these times nectar levels in flowers were often much higher than later in the day and so honeyeaters gained a dis-proportionate share of the resources relative to their daily visitation rates. Paton (1990) also considered the fate of pollen at some of these plants and found that honeybees removed 39-99% of the pollen being produced (table 7).

Similar calculations are still to be made for a range of insect-pollinated flowers. However, the proportion of resources consumed by honeybees will be similar to or higher than the proportion of visits being made by honeybees to the insect-pollinated flowers listed in table 6 for two reasons. First, honeybees usually begin foraging 1-2 hours earlier than native insects at least in temperate locations (eg Paton 1993) and so have more or less exclusive use of these flowers at times when nectar and pollen availability are highest (Paton unpubl.). Second, honeybees being larger than most of the native insects visiting these flowers, usually remove more nectar or pollen during a visit to a flower than the native insects (Paton unpubl.). Given that honeybees often accounted for 80% or more of the visits to flowers of a wide range of insectpollinated plant taxa (table 6) their share of the floral resources will be at least this high. Therefore, interactions between honeybees and Australian biota are not trivial and the potential for competitive interactions is high. These measurements alone, however, do not provide evidence of competition, since honeybees could still be removing floral resources that were not needed by native fauna.

Table 7
Quantities of nectar and pollen removed by honeybees and native fauna visiting plants near Rocky River in Flinders Chase NP, SA. Sets of data illustrate temporal variation in resource use. HB = honeybees; BD = birds; NB = native bees; % resource removed = percentage of nectar and pollen produced that was taken by each taxon; + _ < 0.001. Further details in Paton (1990).
Plant species Months HB resource removed
Eucalyptus cosmophylla
mid Aug 87
E. cosmophylla
late Aug 87
E. remota
Jan 89
Callistemon rugulosus
Nov 88
C. rugulosus
Dec 88
Adenanthos terminalis
Aug 87
A. terminalis
Jan 89
Correa reflexa
May 87
C. reflexa
Jul 87
C. reflexa
Aug 87
E. remota
Jan 89
A. terminalis
Aug 87
A. terminalis
Jan 89
Bond and Brown (1979) estimated honeybees consumed 13-20% of the nectar produced by E. incrassata at Wyperfeld NP, Vic in Oct-Nov 1977 and Collins et al. (1984a) estimated honeybees consumed 39% of the nectar produced by Calothamnus quadrifidus in Nov 1983 at Wongamine Nature Reserve, WA. Honeybees also consumed 36% of the nectar produced by Grevillea aquifolia at Golton Vale, central Vic in Oct 1977, 52% of the nectar produced by Callistemon rugulosus at Golton Vale in Nov 1977, and 34% of the nectar produced by Amyema pendulumat Cranbourne, Vic in Nov 1977 (Paton 1979, 1985). In each example honeyeaters either consumed or were assumed to consume the remainder.


Competition between honeybees and native fauna for floral resources

A number of studies have attempted to show experimentally that honeybees compete with some native fauna for floral resources. These experiments have involved manipulating the numbers of honeybees in an area-usually by adding beehives to a site-and measuring responses of selected native fauna-usually native bees or honeyeaters.

Effects of honeybees on numbers of native bees

Pyke and Balzer (1985) conducted the first manipulative studies. In 1981 and 1982 they manipulated the numbers of honeybees working flowers used by native bees and recorded responses of native bees to these manipulations. In their first experiment they selected eight 1 m2 plots of Leptospermum squarrosum.Four of these plots acted as controls and the other four were' experimental plots. At the experimental plots honeybees were removed on six days between 1 and 17 April 1981. On these days they repeatedly counted the numbers of honeybees and native bees arriving at these census plots for ten minute periods. About four counts/plot/day were made. These counts were then compared with counts collected at the same plots on four days when honeybees were not removed. Both sets of data were also compared with similar data collected at the four control plots where honeybees were not removed on any of the days. More native bees arrived at the experimental plots for those days when honeybees were being removed (table 8) and Pyke and Balzer interpreted this response as indicating a competitive interaction. However, there were still large numbers of honeybees arriving at the experimental plots on these days and the rates of arrival were similar to the numbers arriving at control plots (table 8). Pyke and Balzer (1985) did not state whether the honeybees that arrived during counts at experimental plots on days when honeybees were to be removed were allowed to forage. However, elsewhere in their report (see table 5.1 in Pyke and Balzer 1985), they recorded 28 the time taken by honeybees to visit flowers within these experimental plots on removal days, so at least a proportion of these honeybees must have foraged. The availability of floral resources may not have been very different in the experimental and control areas, since the proportion of floral resources removed by honeybees is not necessarily proportional to the numbers of honeybees foraging in an area (eg Paton 1990).

Table 8
Mean arrival rates of honeybees and native bees per hour at 1 m2 plots of Leptospermum squarrosum on days when honeybees were removed (-HB) and not removed (+HB) from experimental plots (Pyke and Balzer 1985). Standard errors were approximately 10% of the means for honeybees (1.8 to 3.2) and 15-50% of the means for native bees (0.8 to 1 .5).
  Honeybees Native bees
  +HB -HB +HB -HB
Experimental plots
Control plots

An implicit assumption in this study is that native bees can detect differences in the availability of floral resources at a scale of 1 m2 . If they are unable to detect such differences then there is no a priori reason to expect their numbers to increase in experimental plots when some of the honeybees have been removed. Other factors like the densities of flowers in the various plots and adjacent to them rather than the numbers of honeybees arriving at the plots may have influenced the arrivals of native bees. Unfortunately information on the numbers of flowers at control and experimental sites were not given although these were counted at least for the first day. Other pertinent details were also not given. For example, no, information was given on the numbers of honeybees that were removed from each plot, the timing of this removal relative to the counts, or the spatial arrangement of the eight plots although these were all within an area of 100 m2. Given the proximity of the experimental and control plots, removal of honeybees from experimental plots may have also influenced the numbers using nearby control plots. Furthermore, the numbers of arrivals of bees may not be a good measure of insect activity at flowers within the plots. Ideally the length of each visit (ie numbers of flowers probed within the 1 m2 plot) not just the number of visits (arrivals) should be recorded. Low rates of arrival may have been countered by lengthy stays. Without some of this additional information interpreting an apparent increase in the numbers of native bees at the 1 m2 experimental plots in response to removing some of the honeybees as evidence of competition between honeybees and native bees is difficult.

Bailey (1994) reports more convincing results following similar removal experiments conducted near Jandakot, WA during autumn when floral resources were scarce. In the study, the numbers of insects attending flowering Leucopogon propinquus plants were counted for repeated 30 min periods between 1130 am and 1515 pm for three days when honeybees had access and for three days when honeybees were being removed. On days when feral honeybees were being removed the numbers of two species of native bee, Campsomeris sp.and Nomia sp.increased as did an unknown species of Diptera. Other native insects were also recorded but their abundances were too small and. variable to detect any significant changes following removal of honeybees.

Pyke and Balzer (1985) also counted the numbers of honeybees and native bees arriving at 1 m2 plots of Angophora hispida at four distances away from a permanent apiary of 60 hives in Royal National Park near Sydney. They suggested that a decrease in honeybee density (arrivals), and an increase in native bee density, with increasing distance from the apiary would be indicative of a competitive interaction between honeybees and native bees. Such a pattern was not found. The numbers of both honeybees and native bees visiting plots of Angophora flowers were highest at the furthest distance (800 m) from the apiary at which Pyke and Balzer collected data. The notion that there should be a decline in honeybee activity with distance from an apiary may not always hold. For example, Visscher and Seeley (1982) found that foraging honeybees were often patchily distributed by distance and direction from their hives and that the dispersion patterns of foraging honeybees often changed dramatically from one time period to the next. In addition the densities of honeybees foraging in different plots in Royal NP may have been influenced by the distribution and abundance of feral colonies. Honeybees Native bees

Pyke and Balzer (1985) also introduced 30 hives of honeybees to an alpine area of Kosciusko NP where there had been no previous history of honeybees. Responses of native bees to this introduction were measured at four distances from the apiary by counting the numbers of bees seen along transects and the numbers arriving at 1 m2 census plots of flowering Prostanthera cuneata before and after the introduction of the hives. The two sets of data provided some interesting contrasts. Both transect data and arrival data showed that honeybee densities were greatest at sites close to the apiary than at sites further away (table 9). Transect counts also showed that native bee densities were consistently higher close to the apiary (50 m) than further away both before and after the arrival of honeybees (table 9). At sites 1000 m from the apiary the densities of native bees remained low and did not change dramatically following the introduction of honeybees. At sites closer to the' apiary the densities of native bees declined following the introduction of honeybees. At sites 50 m from the apiary the density was halved but at sites 200 m and 400 m from the apiary densities of native bees were only 10-20% of the prehoneybee counts. If honeybees were competing with native bees for floral resources then the declines should have been more severe at sites closest to the apiary where honeybee densities and activity was highest (approximately 4 times higher than at other distances).

Counts of native bees arriving at census plots, however, did not mirror counts of native bees along transects (table 9). No explanation was given for these differences. Although counts of native bees arriving at census plots declined close to the apiary they still remained high relative to areas further from the apiary. If honeybees were competing with native bees for resources then native bees might be expected to avoid areas with the highest honeybee densities. This was clearly not the case.

The inconsistencies between the two methods for censusing native bees weakens Pyke and Balzer's claim that these experimental data provide evidence of competition. A variety of abiotie factors (eg temperature, wind, time of day, cloudiness) are likely to influence foraging activities of native bees (eg Paton 1993) and ideally simultaneous counts should have been made in the various areas to allow valid comparisons. Pyke and Balzer (1985) gave no details on when different areas were sampled with each of the two techniques or whether each area was sampled under comparable weather conditions. Slight differences in ambient conditions from one day to the next, from one location to another (eg east- and west-facing slopes) and from one time interval to the next within a day could have influenced the activities of native bees when counts were made, and so produced the patchy and inconsistent sets of results. Pyke and Balzer unfortunately did not consider these possibilities when designing their sampling program and only censused bees in one direction away from the apiary.

Other difficulties potentially exist with these data. For example, Pyke and Balzer (1985) provided few details on the species of native bees and other taxa that were recorded in plots or along transects before and after the honeybee manipulations and how the counts for each species of native bee and other taxa varied spatially and temporally. Identification to species, however, is often difficult without microscopic examination of the individual bees and so that is an understandable omission.

In conclusion, Pyke and Balzer (1985) have not provided convincing evidence of competitive interactions between honeybees and native bees but nor have they shown that competitive interactions are non-existent. Rather their work illustrates the paucity of good ecological data on the foraging ecology of Australian native bees and highlights the complexity of interactions between honeybees and native bees. It is a starting point for further studies.

This lack of knowledge and complexity is easily illustrated. For example, the usual assumption is that if honeybees compete with native bees for floral resources then the numbers of native bees counted at a patch of flowers should decline following the introduction of honeybees (Pyke and Balzer 1985; Pyke 1990); honeybees simply displace native bees from flowers. Paton (1993), however, suggested that in a competitive environment the numbers of native bees recorded foraging in an area might actually increase rather than decrease following the introduction of honeybees, reflecting a reduction in the quantities of floral resources encountered at flowers.

This may be the scenario. Female native bees collect nectar and pollen, and package that food with an egg in a chamber or cell within a nesting burrow and then repeat this procedure (Michener 1970). Assume that there are 100 female bees in an area and that the bees take 5 minutes to collect a load of nectar or pollen before returning to their burrows. Once in the burrow, they take 10 minutes to unload before going out to forage again. Now introduce honeybees that remove a large proportion of the nectar and pollen produced in the area. As a result, native bees find less food at each of the flowers they visit and consequently spend more time foraging to obtain a full load. To simplify the calculations assume that under the new conditions with honeybees present a native bee takes twice as long (10 minutes) to collect a load before returning to her burrow. Because unloading still takes 10 minutes, the net effect of introducing honeybees might be to increase the numbers of native bees working flowers at one time. In the example given here the numbers of native bees foraging simultaneously would have risen from 33 to 50, despite the actual numbers living in the area remaining at 100. Thus changes in the numbers of native bees counted at patches of flowers following the introduction of honeybees are not easily interpreted, except that changing numbers indicate some interaction.

Table 9
Numbers of honeybees and native bees counted along 10 m sections of transects and in 1 m2 plots of Prostantheracuneataat different distances from an apiary of 30 hives stationed in Kosciusko NP. Data are from before and after the introduction of the hives (Pyke and Balzer 1985) .
  Honeybees Native bees
  transects plots transects plots
Distance from apiary (m)
before after before after before after before after

Effects of honeybees on the reproductive performances of native bees

Ultimately impacts of honeybees on native bees should result in long term reductions in the population densities of native bees. Three studies have attempted to measure the influence of honeybees on the reproductive performances of some species of native bees. All three studies have involved adding hives to areas where there were also background feral colonies followed by measuring the reproductive performances of a particular species of native bee in areas with and without these added hives. Only one of these studies has been published.

Sugden and Pyke (1991) measured the population biology and reproductive performance of Exoneura asimillima in areas where they had introduced beehives and compared these to similar measurements made in control areas 7-8 km away (outside the flight range of the introduced hives). In the first year of measurements the only differences were significantly more large larvae and prepupae within the colonies near the apiary than at the control sites and this difference was dismissed as being a possible artefact of sampling colonies near the apiary two weeks after those at control areas. As a result colonies near the apiary may have been more advanced and so had more larvae and prepupae. In the second year more founder colonies established near the apiary and these had more eggs, larvae and pre-pupae than founder colonies at the control sites (table 10). Survival of both established and founder colonies was similar at both control and experimental sites. However, there were fewer adults of both sexes in the established colonies at the experimental sites (near apiary) compared with the control sites (table 10).

These data are difficult to interpret, particularly given that possible differences in floral resources and microchmates may have existed between sites, and accounted for differences in the composition of colonies between sites (Sugden and Pyke 1991). Unfortunately these experiments were not replicated, and although there were eight experimental plots these were all at the one experimental site and so these other factors cannot be eliminated. Information on the variability between plots in the composition of colonies is also not given. Nevertheless, the authors interpreted these data to imply a negative effect of honeybees on local colonies of E. asimillima.This conclusion was based largely on the lower numbers of adults remaining at established colonies near the apiary.They suggested that competition for food, especially nectar, was possible since honeybees altered the foraging patterns of E. asimillima but provided no details. Despite the lower numbers of adults remaining at established colonies, these colonies and founder colonies were at least as fecund if not more so near the apiary than they were at control sites further away and at a larger number of new colonies founded in the experimental sites. Such a result is not consistent with the notion that the bees were competing with honeybees for floral resources. Other possibilities suggested by Sugden and Pyke (1991) were second order interactions where honeybees affected the foraging behaviour, of predatory ants and/or parasites of E. asimillima which in turn influenced Exoneura colonies.

Two other studies have been measuring the effect of honeybees on the reproductive biology of native bees. Schwarz, Kukuk and Gross (pers. comm.) measured the reproductive performances of Exoneura bicolor in control and experimental areas in Cobboboonee State Forest near Portland, Victoria. Four control and four experimental areas were established with experimental areas receiving additional hives of honeybees. These areas had a low background level of feral honeybees (M Schwarz pers. comm.). Initial measurements have revealed that colony survival and brood production were higher in areas where hives had been placed. This result is not consistent with a direct competitive effect of honeybees on Exoneura.Kukuk and Schwarz have suggested three possible explanations for these results:

  1. subtle differences in resource availability between sites with and without added hives of honeybees such that sites with additional honeybees were still better for Exoneura;
  2. high honeybee densities leading to the competitive exclusion of some other floral visitor(s) which would otherwise have appropriated a large proportion of the floral resources, leaving more food available for Exoneura; and
  3. high honeybee densities leading to satiation or prey specialisation by insectivores that would otherwise have preyed on Exoneura.

Various predatory ants (eg species of Myrmecia) are often seen amongst the flowers of plants waiting to grab unsuspecting insects probing flowers (D Paton pers. obs.; M Schwarz pers. comm.). In Flinders Chase NP most of the insects caught amongst flowers by these ants were honeybees that responded more slowly than native bees to an approaching ant (D Paton pers. obs.). Spessa and Schwarz (pers. comm.) are also measuring the reproductive performances of a native colletid bee Amphylaeus morosus at sites with and without added beehives in the Black Ranges and Toolangi State Forests, Victoria. Four control and four experimental sites were established in these montane forests with each of the experimental sites receiving 6 hives of honeybees. The reproductive performances of the colletid bee was measured in these plots over two seasons. Preliminary analyses suggest that there is no conspicuous impact of honeybees on this bee, though pupal masses maybe slightly, lower at experimental sites (A Spessa pers. comm.). Other explanations for these lower weights, however, still need to be explored.

In each of these studies, the introduction of hives of honeybees is assumed to reduce the availability of floral resources at experimental sites relative to the control sites. None of the studies, however, measured resource abundance at control and experimental sites in sufficient detail to determine if resource availability was reduced following the introduction of hives. This information is needed to properly interpret responses of native bees. If resources declined yet the reproductive outputs of native bees were not affected then honeybees were not competing with native bees. On the other hand, if the availability of floral resources did not change following the introduction of additional hives of honeybees then concluding that no competition exists when the reproductive outputs of native bees at control and experimental sites were the same may be incorrect for two reasons. First, simply introducing hives of honeybees does not guarantee that there will be an increase in honeybee activity at the flowers being used by the native bees being studied at the experimental sites. Honeybees from hives will spread their foraging effort over a large area; conservatively estimated at 12 km² (Visscher and Seeley 1982; Winston 1987). Thus any effect that the additional honeybees may be having on floral resources may be dissipated over a large area and have a relatively small effect on native bees in any one area that is difficult to detect. Furthermore, although control sites may be beyond the normal flight ranges of honeybees stationed at an experimental site, their presence at flowers in other areas may force honeybees from other (feral) colonies to re-distribute their foraging effort such that the amount of foraging done by honeybees in control areas also increases. Second, increases in the numbers of honeybees working flowers does not necessarily mean that honeybees will have a larger share of the resources. Paton (1990) showed that increases in the numbers of honeybees working flowers had a diminishing effect on the proportion of floral resources that they consumed; at relatively high densities of honeybees the addition of further honeybees did not increase their share of the resources. In each of the above studies there were background levels of feral honeybees and these may have already been consuming most of the available resources; introducing additional hives may have had no added effect on the resources available to native bees. In future, measuring the responses of native bees to the removal of feral colonies of honeybees from an extensive area may be a more appropriate manipulation than introducing additional hives of honeybees.

Table 10
Composition of established and founder nests of Exoneura asimillima at control plots and experimental plots in Nadgee Nature Reserve, NSW in Feb 1987.Sugden and Pyke (1991) removed all the Xanthorrhoea scapes from each of eight 10 x 10 m plots at an experimental site and eight similar sized plots at control sites. To each of these plots they added approximately 10 active nests and 44 blank scapes which served as potential nest sites for founder colonies on or before 16 Nov 1986. All nests were then collected on 15 Feb 1987 and their contents scored. Hives of honeybees were stationed near the experimental plots during the experiment. Initially 29 hives were introduced on 12 Nov 1986; some died; weak colonies were combined; the 17 active colonies remained when the hives were removed on 28 March 1987 (Sugden and Pyke 1991).Figures are the average number of eggs, larvae, pre-pupae and adults in nests taken from different sites.
  Founder nests Established nests
  Experimental Control Experimental Control
No. of colonies (n)
No. of adult males
No. of adult females
Large larvae and pre-pupae
Eggs and small larvae
* significantly greater than the equivalent data for control or experimental plots (p<0.05 at least) totals as given in Pyke and Sugden (1991), not equal to the sum of eggs, larvae, pre-pupae and adults


Effects of honeybees on the foraging behaviour and abundance of nectar- feeding birds

In Australia, more than 100 species of birds have been seen harvesting nectar from flowers (Ford et al. 1979). Most of these are species of honeyeater (Meliphagidae), some of which depend on nectar or similar carbohydrates for energy (Paton 1980, 1982a, 1988). Numbers of honeyeaters living in areas in southern Australia are often correlated with the quantity of nectar being produced, and breeding usually coincides with periods of abundant nectar (Ford 1979; Ford and Paton 1985; Paton 1979, 1985; Pyke 1983, 1988; Pyke and Recher 1986). In some cases, the birds defend clearly defined feeding territories in which dominant individuals aggressively exclude intruders to gain more or less exclusive use of nectar in their territories (Ford 1981; Paton 1979, 1985, 1986; McFarland 1986c; Newland and Wooller 1986). Although honeyeaters can defend floral resources from subordinates they cannot prevent honeybees from removing a substantial share of the resources (table 7; Paton 1979, 1985). Given the importance of nectar to these birds such losses of nectar to honeybees are likely to affect the abundance and behaviour of the birds.

Only Paton and co-workers (1993, unpubl.) have experimentally manipulated honeybee numbers and measured the responses of birds to these manipulations. Two studies have been executed. The first involved documenting the behavioural responses of New Holland Honeyeaters, Phylidonyris novaehollandiae to changes in the numbers of honeybees foraging at flowers of Caltistemon rugulosus during spring at Scott Conservation Park, near Goolwa, South Australia. The second involved measuring the numerical responses of honeyeaters to the introduction of commercial loads of honeybees to Banksia ornata heathlands during winter in Ngarkat Conservation Park. In the Callistemon study all of the nectar being produced by the plants was being consumed by floral visitors, predominantly New Holland Honeyeaters and/or honeybees. The numbers of honeybees working the flowers increased: with proximity to a large apiary on the western boundary of the park; seasonally as honeybees switched to Callistemon flowers as other resources declined; and after the introduction of ten additional hives to an experimental plot. Honeyeaters responded to increases in the numbers of honeybees working Callistemon flowers by reducing the frequency with which they visited individual flowers. For example, when few honeybees worked the flowers, New Holland Honeyeaters visited individual flowers on average 9.6 times/ day, but when honeybee activity was high this visitation was reduced significantly to only 3.0 visits/flower/day (Paton 1993; table 11). In addition the birds adjusted their foraging by avoiding the flowers that were most extensively used by nectar-feeding honeybees.

Honeybees showed distinct preferences for certain flowers when harvesting nectar from this Callistemon. For example, they visited flowers at the two ends of an inflorescence more frequently than those in the centre (table 12). They also favoured inflorescences exposed on the ends of branches at the periphery of the plant's canopy over those that were completely or partially hidden by foliage within the plant (table 12). These patterns of use by honeybees are easily explained by the ease with which different flowers can be visited by honeybees (see Paton 1993 for details).

New Holland Honeyeaters responded to these patterns. In the absence of honeybees, New Holland Honeyeaters showed no patterns to floral visitation, visiting all flowers equally (table 13). However, when honeybees were working the flowers, the birds showed a strong bias to inflorescences borne deep within the canopy of a plant and also a strong bias to visit the centrally-located flowers within an inflorescence (table 13). On average these inflorescences and flowers had higher standing crops of nectar than the flowers and inflorescences being used extensively by honeybees (Paton unpubl. data). Clearly, honeybees altered the foraging patterns of the birds, with the birds concentrating their foraging activity at the flowers used least by honeybees.

Table 11
Changes in the frequency with which New Holland Honeyeaters visited the flowers of Callistemon rugulosus and changes in territory sizes for these birds with changes in the numbers of honeybees working the flowers near Goolwa, SA in spring 1983 .Honeybee activity was scored by counting honeybees at 50-200 inflorescences at regular intervals through the day and converting these to bees per 1000 flowers. Daily visitation by birds involved counting the number of visits made by birds to samples of flowers during five 1 hour periods spread evenly over a 14 h day. Data are expressed as mean ± se (n), where n for honeyeater visits is the number of independent determinations made during the level of honeybee activity, and n for territories is the number of territories measured. Visitation rates by birds to flowers were significantly lower (ANOVA, F = 12.7, df = 3,32, p< 0.001), and territory sizes significantly larger (ANOVA, F = 21.3, df = 3,106, p<0.001) when honeybee numbers were higher.
Maximum no. of honeybees seen per 1000 flowers Honeyeater visits/flower/day No. of flowers in honeyeater territories
9.6 ± 0.8 (9)
4343 ± 185 (16)
5.5 ± 1.0 (9)
5387 ± 219 (38)
5.0 ± 0.8 (9)
6112 ± 306 (28)
> 15.0
3.0 ± 0.3 (9)
7606 ± 304 (28)

The presence of honeybees working the flowers may also influence diurnal patterns of foraging by honeyeaters. When honeybees were present in only small numbers, New Holland Honeyeaters foraged more or less equally throughout the day. However, when honeybees were frequent visitors to flowers, honeyeaters foraged more intensively early in the morning and less so during the middle of the day when honeybees were most active (Figure 1). Thus losses of nectar to honeybees during the middle of the day may also affect diurnal patterns of foraging in the birds.

Bond and Brown (1979) showed similar diurnal patterns of activity for honeyeaters and honeybees foraging on E. incrassata nectar, as did Paton (1979,1985) for New Holland Honeyeaters and honeybees visiting a range of other plant species. Stace (1988) subsequently concluded that this habit of birds foraging intensively early in the morning, with honeybees foraging in the middle of the day, eliminated much of the potential competition between honeyeaters and honeybees. However, Stace (1988) did not consider the possibility that foraging by honeybees in the middle of the day influenced when the birds foraged. Another assumption was that honeyeaters could harvest much of their daily food requirements during the first few hours of the day and store this food for later use. Initial research, however, shows that honeyeaters may have a limited capacity to rapidly store energy, and when given the opportunity prefer to steadily accumulate food reserves throughout the day (eg Collins and Clow 1978; Collins and Morellini 1979; Collins et at. 1980). In any case a moderate level of foraging continues throughout the day even in the presence of honeybees (Paton 1979, 1982a, 1985; Bond and Brown 1979; Figure 1).

Table 12
Spatial patterns in the use of Callistemon rugulosus flowers by nectar-collecting honeybees. Use of inflorescences on a plant was scored by counting the numbers of honeybees foraging at 250 inflorescences fully exposed on the exterior of the plant's canopy, 250 partially covered and 250 fully covered by the plant's foliage. Use of flowers on an inflorescence was scored by dividing the inflorescence into three equal parts (proximal, middle and distal third) and counting the numbers of honeybees foraging at flowers in each third. Honeybees used exposed inflorescences more extensively than those that were partially or fully covered (X2 = 26.4, df = 2, p<0.001) and used flowers in the distal and proximal thirds of an inflorescence more than flowers centrally located within an inflorescence (X2 = 25.8, df = 2, p<0.001).
Position of inflorescence on plant No. of honeybees counted Position of flower on inflorescence No. of honeybees counted
proximal third
partially covered
middle third
fully covered
distal third
Table 13
Spatial patterns to the use of Callistemon rugulosus flowers by New Holland Honeyeaters in the presence and absence of honeybees. Use of inflorescences was scored by recording the frequency with which honeyeaters visited inflorescences that were exposed, partially covered or fully covered by the plant's canopy during one hour observations. Exposed inflorescences were more numerous than partially covered and fully covered inflorescences, so data are expressed as visits per inflorescence per hour to allow easier comparisons. The number of inflorescence hours are given in parentheses. Use of flowers within an inflorescence was scored by recording the numbers of probes made by New Holland Honeyeaters at flowers in the proximal, middle and distal thirds of inflorescences. 550 probes were scored at times when honeybees were not foraging and 1346 probes were scored when honeybees were foraging. New Holland Honeyeaters used all inflorescences and all flowers equally when honeybees were absent (X2 = 2.59 and 3.00 respectively, df = 2, p>0.05) but favoured covered inflorescences and centrally-located flowers within an inflorescence when honeybees were present (X2 = 727 and 133 respectively, df = 2, p<0.001).
  Frequency of use by New Holland Honeyeaters
Position of flower or inflorescence Honeybees absent Honeybees present
3.44 (133)
1.26 (848)
partially covered
3.68 (57)
2.62 (424)
fully covered
4.06 (193)
% of 550 probes
% of 1346 probes
proximal third
middle third
distal third

These responses of honeyeaters to losses of food to honeybees are consistent with a competitive interaction but they are not sufficient on their own to indicate that honeybees were causing population densities of honeyeaters to decline. Paton (1993), however, also examined the effect of nectar losses to honeybees on the sizes of the territories being defended by New Holland Honeyeaters using C. rugulosus. If honeybees have no influence on the ability of the birds to harvest nectar, then changes in the numbers of honeyeaters working flowers should not alter the sizes of the birds' territories. However, the number of Callistemon flowers defended by territorial New Holland Honeyeaters increased significantly when the numbers of honeybees working the flowers increased (table 11). When large numbers of honeybees were visiting the flowers, New Holland Honeyeaters defended more than 7000 flowers, almost double the number being defended when only a few honeybees were working the flowers (table 11). This doubling of territory size was consistent with the amounts of nectar (ca 50% of production) calculated to be lost when large numbers of honeybees were foraging (ie > 15 honeybees/ 1000 flowers) at Callistemon flowers.

This increase in territory size with increases in honeybee activity was also supported with some experimental evidence (Paton 1993). When the number of honeybees working flowers was increased experimentally by placing 10 hives next to a patch of Callistemon, the dominant birds (adult males) in that patch expanded their territories by displacing subordinates (juveniles and females) from adjacent territories and adding all or parts of these territories to their own. In the experimental patch the territories of five dominant birds increased significantly from 4744 + 216 (se) to 7523 + 290 (se) flowers after the introduction of the bee hives (ANOVA, p<0. 001, Paton 1993; table 14). This increase in territory size was consistent with. changes of from around 10% to approximately 50% of food lost to honeybees following the. introduction of the beehives. In control areas where honeybee numbers did not change, territories of five individual birds did not increase significantly (table 14). Note, however, that the responses of individual birds within a patch are not strictly independent.

Figure 1
Figure 1 Diurnal changes in the percentage time spent foraging by individual New Holland Honeyeaters when the numbers of bees working C. rugulosus flowers were low (dashed lines) and when they were high (solid lines). Each point is the percentage time of a 1 hour time budget spent foraging. Data collected Oct-De, 1983 at Scott CP, SA.

Although this manipulation of honeybee numbers and subsequent territorial responses of honeyeaters needs to be replicated, the results of this trial are consistent with the other behavioural responses that were recorded following changes in honeybee activity at Callistemon flowers. The results suggest that the numbers of honeyeaters living in an area could be reduced to about a half of the carrying capacity expected if honeybees were absent. Furthermore, females appear to be displaced more frequently than males and this disproportionate loss of females may affect honeyeater population dynamics more than if both sexes were displaced equally (Paton 1993).

Paton (1979, 1985) also showed that territorial New Holland Honeyeaters held larger feeding territories on a range of other plants when honeybees were also harvesting the floral resources from these plants. Again the increase in the size of the territories (numbers of flowers defended) matched the amounts being consumed by honeybees such that the birds still defended sufficient resources to meet their energy requirements. These results are consistent with honeybees competitively excluding some birds from flowers but they contrast dramatically with studies on changes in the numbers of honeyeaters using Banksia ornata heathlands during winter in Ngarkat CP following the introduction of beehives.

Ngarkat CP is an important overwintering site for commercially-managed honeybees in South Australia. In all there are over 200 registered apiary sites within this reserve, though typically only about 80 sites are stocked with honeybees in any one year. Most of these sites are in the western and southern sections of the reserve with beekeepers shifting hives into the area in late May and maintaining them in the reserve until late July or early August. The primary plant species providing floral resources during this period is the desert banksia, Banksia ornata and in most years native animals are unable to prevent floral resources (both nectar and pollen) from accumulating at these inflorescences (Paton et at.unpubl.).

Table 14
Changes in the sizes of territories of New Holland Honeyeaters following the introduction of 10 hives of honeybees to an experimental site. Upon the introduction of the hives the frequency with which honeybees visited C. rugulosus flowers increased from 0-8 visits/flower/day to 36-44 visits/flower/day. On the control sites visitation rates also increased but not to the same extent; from 0-5 visits/flower/day to 16-21 visits/flower/day. These changes in honeybee numbers resulted in food losses increasing from 0-10% to 40-50% after the extra hives arrived at the experimental site. At the control site losses to honeybees increased from 0-10% to 20-25%. Individually colour-banded honeyeaters were watched for up to 5 h to determine the boundaries of their territories. The numbers of inflorescences present in each territory was then counted and the numbers of flowers calculated by multiplying the number of inflorescences by the mean number of flowers counted on 50 inflorescences selected within each of the territories.
    Territory size (no. of flowers)
Bird# Before honeybees after honeybees
Experimental site
Control sites

Since 1990; Paton and co-workers have been assessing the impact of commercial beekeeping operations on the flora and fauna of Ngarkat CP to help land managers decide whether to allow continued use of the reserve by beekeepers. The research involved selecting 15 sites (each at least 3 km away from any other site) within the central part of the reserve. Most of this area had had no previous history of commercially-managed honeybees and densities of feral colonies were negligible (0.001 col/ha). Some of the sites and not others then received commercial loads of honeybees in one or more seasons and the responses of native biota to those manipulations measured. Amongst the taxa considered were nectar-feeding birds, small mammals, native bees and several other groups of flower-visiting insects, including ants and staphylinid beetles. The research also involved measuring the production and availability of floral resources and seed production by the plants.

Although the presence of honeybees reduced the quantities of nectar available at Banksia inflorescences, particularly near apiaries, there were still considerable quantities of nectar remaining at the end of the day when honeybee foraging had ceased (table 15; Paton et at.MS3; unpubl.). The quantities of nectar left over often exceeded 0.5 g of sugar/ inflorescence even within 100 m of an apiary. These leftover quantities were still more than adequate to satisfy native fauna. For example, a typical 20 g honeyeater needs about 5 g of sugar per day to satisfy its energy requirements (eg Paton 1982a). So even within 100 m of an apiary the bird still only needed to visit 10 inflorescences at the end of the day to collect its daily energy requirements. In these areas there can be 1000 inflorescences in bloom/hectare so even after the birds and bees had fed during the day there was still enough food left over to feed the equivalent of another 100 birds/ha.

Consistent with this surfeit of food there were no significant differences in the numbers of honeyeaters counted at sites stocked and not stocked with honeybees (table 16). Nor were there any significant differences between sites with and without honeybees in the numbers of small nectar-feeding mammals caught in pitfalls or invertebrates counted at inflorescences (Paton et at.MS3). So surplus floral resources existed in this system and although honeybees depressed food availability, the level of depression was not sufficient to affect the abundance of native fauna in these heathlands. One explanation for the existence of surpluses is that there were insufficient native fauna, particularly birds, at Ngarkat to fully exploit floral resources. In other B. ornata healthlands closer to the coasts of South Australia, densities of honeyeaters ranged from 6-22.5 honeyeaters/ha when B. ornata was flowering, two to eight times the density recorded at Ngarkat (Paton unpubl.). Honeybees were also prominent at inflorescences in these areas. One reason why there might be relatively few birds in Ngarkat during winter when B. ornata flowers is because this area produces few suitable flowers for these birds during summer and autumn and so many of the birds must leave the park for that period. The areas that will produce suitable resources for these birds during summer and autumn are the more mesic coastal and woodland areas. These areas, however, have been extensively and disproportionately cleared for agriculture compared to the drier sandy areas like Ngarkat that are least suited to agriculture. As a result, population- sizes of honeyeaters in the region as a whole may be severely limited by the availability of nectar sources during the summer and autumn months. Thus honeyeaters cannot recruit back into the drier heathland habitats of Ngarkat in sufficient numbers to fully exploit floral resources during winter. Such an explanation is consistent with seasonal patterns to nectar availability that have been reported from other areas in general (see above) where summer sources are more heavily exploited than winter sources in many areas. Furthermore, honeybees usually consume a larger share of the floral resources being produced during the warmer months of the year and may depress resources further for honeyeaters at these times. The irony of this scenario is that experiments to measure the impact of honeybees on the flora and fauna of Ngarkat CP should probably have involved manipulating honeybee numbers in areas outside the park and at other times in the year rather than manipulating the numbers of honeybees using this reserve in winter when B. ornata was flowering.

Table 15
Quantities of nectar remaining at Banksia ornata inflorescences in late afternoon at sites stocked and not stocked with honeybees in Ngarkat CP during the winter of 1990. Inflorescences were collected at approximately 1600h, after honeybee activity had ceased for the day. This was accomplished by carefully placing a plastic bag over the inflorescence and snapping off the inflorescence so that any dislodged nectar would be collected in the bag. Nectar was removed using the centrifuging technique of Armstrong and Paton (1990) and the volume and concentration measured and used to calculate the grams of sugar present in the nectar. Nine sites were sampled in each month (3 stocked with honeybees and 6 without honeybees). For these monthly samples 12 inflorescences were collected at 100 m and 12 inflorescences at 1000 m from the central point of each site; 6 inflorescences being collected at each distance along each of the two transects established for bird censusing (see table 16). The mean quantity of sugar per inflorescence ± se is given in the table, where n is the number of inflorescences sampled. Almost identical patterns were found in 1992 and 1993 (not shown). No bees were placed in the park in 1991. Although the quantities of nectar remaining at inflorescences near apiaries is lower than control sites and lower than for inflorescences 1 km from the apiary, there are still substantial quantities of nectar left unexploited by honeybees near apiaries.
  Sugar (g) per inflorescence
sites with honeybees sites without honeybees
100 m from the central point of each site
1.15 ± 0.04 (36)
1.52 ± 0.07 (72)
0.59 ± 0.20 (36)
1.35 ± 0.39 (72)
0.49 ± 0.14 (36)
0.70 ± 0.14 (72)
1000 m from the central point of each site
1.40 ± 0.26(36)
1.62 ± 0.02 (72)
1.22 ± 0.32 (36)
1.10 ± 0.27 (72)
0.86 ± 0.05 (36)
0.66 ± 0.17 (72)
Table 16
Densities of honeyeaters (no./ha) at sites stocked and not stocked with commercial loads of honeybees in Ngarkat CP, SA. Honeyeaters were counted along two 1500 m transects radiating out from the central point of each of the 15 sites. Each transect was marked with flagging tape, at 50 m and 100 m intervals. Observers walked these transects at the rate of 4 minutes/100 m commencing at 8 am. All birds seen within 50 m of the transect line were recorded for each 100 m section of the transect. Birds that flew over were also noted. Birds along each transect were counted twice during a census, once on the way out and again on the way in. In the following table, densities of honeyeaters are shown for the first 500 m of transects (from the central point). Data from separate sites have been pooled for ease of presentation. In 1990 and 1992 five of the fifteen sites were stocked with honeybees while in 1993 nine sites received commercial loads of honeybees. No honeybees were placed on sites in 1991 because of a drought. The table shows the mean ± se number of honeyeaters counted per hectare for sites with and without commercial loads of honeybees. Samples sizes are the number of independent censuses made at the two types of sites. Densities of honeyeaters were not significantly lower at sites stocked with honeybees.
  Number of honeyeaters per hectare
sites without honeybees sites with honeybees
2.89 ± 0.39 (46)
2.92 ± 0.29 (20)
2.26 ± 0.16 (60)
2.21 ± 0.24 (30)
2.29 ± 0.17 (36)
2.61 ± 0.16 (54)


Effects of honeybees on the pollination of Australian plants


Honeybees could alter the pollination rates of Australian plants in many ways. They could:

  1. add to the pollination services provided by native fauna leading to increases in seed production;
  2. displace native pollinators from flowers without providing equivalent pollination services leading to declines in seed production;
  3. alter the behaviour of native pollinators in ways that alter patterns of pollen dispersal leading to changes in seed production; and
  4. remove pollen from flowers reducing the quantities of pollen being transferred to flowers by legitimate pollinators again reducing seed production.

These potential impacts of honeybees on native flora have rarely been considered in Australia.

Matthews (1984) raised two issues concerning interactions between honeybees and native flora: that honeybees were inefficient pollinators of some native plants; and that honeybees may increase rates of hybridisation between plant species. Evidence that hybridisation between coflowering plant species has been facilitated by honeybees, however, is lacking. In fact, many species of native fauna are just as likely to effect interspecific pollen flow as honeybees, if not more so, judging from the mixed pollen loads collected from many native birds, mammals and insects (Paton and Ford 1977; Hopper 1980b; Ford and Paton 1982; Ford and Pursey 1982; Bernhardt et al.1984; Kenrick et al.1987). Honeyeaters also frequently switch from feeding at the flowers of one plant species to another when two or more suitable species are flowering nearby and will defend territories that include more than one species of flowering plant (Paton 1979, unpubl.; Hopper and Burbidge 1986). Under these conditions honeyeaters are likely to effect interspecific pollen flow. Gross (1992) also reports that native bees (Trichocoltetes sp) frequently shifted from one species of pea to another species of pea during foraging bouts if two or more similar species of pea were interspersed. Despite the frequency of interspecific movements by pollinators, the frequency of hybrids is low, suggesting that post-pollination mechanisms in the plants effectively select against interspecific pollen. Even if hybrids were prominent, separating the contribution of honeybees from that of native fauna in the production of these hybrids would be difficult. Moreover, Wapshere (1988) also indicated that the rate of hybridisation amongst Phebalium, a genus of Australian insect pollinated plants, was no greater than the rate of hybridisation amongst Casuarina, a genus of Australian plants usually pollinated by wind.

Most of the concern about honeybees affecting native flora, however, is based on observations that honeybees forage at flowers of some native plants in ways that differ from native pollinators and in ways that were less likely to effect pollination. For example, honeybees rarely touched the stigmatic and pollen-bearing surfaces of several predominantly vertebratepollinated taxa including Grevillea, Banksia, Amyema and Callistemon when harvesting nectar, though pollen-harvesting honeybees visiting these plants made more frequent contact with stigmatic surfaces (Paton and Turner 1985; Paton 1986, 1993, unpubl.; Taylor and Whelan 1988; Vaughton 1992). Bell (1987) also suggested that honeybees robbed nectar or visited the flowers of Templetonia retusa, and species of Crotalaria, Erythrina, Bossiaea, Gastrolobium, Oxylobium, Calothamnus, Banksia and Beaufortia in ways that bypassed the stamens and stigma but provided no quantitative data. Other examples, include Gross's (1993) observations of honeybees contacting the stigmas of the beepollinated Melastome affine less frequently than most of the native bees that visited the flowers. Honeybees also differed from many of the native bees by not buzzing the anthers of this plant to collect pollen. Instead they occasionally harvested pollen that had previously been deposited on stigmas.

Although these examples suggest some inefficiency on the part of honeybees, honeybees can still effect some pollination at many of these plants. For example, Vaughton (1992) showed that inflorescences of Banksia spinulosa that were caged to exclude birds but not honeybees set comparable quantities of seed to open-pollinated flowers where both birds and bees had access, but only late in the season when honeybees were frequent 40 visitors to flowers. Similarly caging experiments showed that honeybees were capable of pollinating Banksia ericifolia, B.ornata, Callistemon rugulosus and Correa reflexa although the quantities of seed produced may be lower than that effected when birds also had access (Paton and Turner 1985; Paton 1993 and see below). Any ineffectiveness in the pollinating abilities of honeybees, however, is unlikely to have a significant negative effect on seed production unless honeybees have displaced a substantial number of native pollinators from the plants or altered the quantities of pollen available for dispersal by native animals.

Paton (1993) considered the impact of honeybees on seed production by Callistemon rugulosus in a small reserve near Goolwa, South Australia. This Callistemon was largely self-incompatible and needed crosspollination to set substantial quantities of fruit. When 685 flowers were cross-pollinated by hand 45.4% set fruit, but only 11.0% of 610 bagged flowers set fruit following selfpollination. Thus to be effective, pollinators should regularly contact the reproductive parts of the flowers and move frequently between plants. Honeybees harvesting nectar from Callistemon, however, only struck the stigma on 4.4% of visits (in over 8000 observations). Pollen-harvesting honeybees struck the stigma more frequently but still on only 16.7% of 1649 visits scored. New Holland Honeyeaters, on the other hand, frequently contacted the stigma of the flower being probed (>50% of occasions) as well as adjacent flowers (determined from photography). Honeybees also rarely moved between plants. For example, in areas where individual plants were widely spaced (>3 m apart) individual honeybees were tracked for a total of 9.9 hours and were observed probing over 4600 flowers.' Not once during these observations was a honeybee observed to fly to an adjacent plant. In these areas territorial New Holland Honeyeaters moved between plants 7.3 times per hour (10 hours of observation), equivalent to one interplant move every 400 probes. Thus when honeybees displace honeyeaters from patches of Callistemon (see tables 11, 14), the quality of the pollination service should decline. If seed production is limited by pollinators then such a displacement of a more effective pollinator by a less effective one should lead to reduced seed production and such a pattern was found. Rates of fruit production for C. rugulosus varied with the numbers of bees and birds working the flowers. First, the numbers of flowers that set fruit inside wire mesh cages, which excluded birds, increased as the numbers of bees increased (table 17) indicating that honeybees could pollinate Callistemon flowers. However, the rates at which caged flowers set fruit (7-17%; table 17) were similar to rates achieved following self-pollination (11%) and well below those achieved after cross-pollination (45%). The low fruit production at caged flowers was therefore consistent with honeybees effecting little cross-pollination for this population of Callistemon. Fruit production at flowers exposed to both birds and bees, however, was significantly higher than that for caged flowers (table 17), indicating that birds provided important pollination services to the plant. Furthermore, this fruit production declined significantly from 35.1% to 22.6% as the numbers of honeybees using the flowers increased (table 17). Thus displacement of pollinating birds by less effective honeybees reduced fruit production for this population of Callistemon.

Honeybees may also alter rates of pollination for Correa reflexa by removing pollen that would otherwise be dispersed by birds (Paton 1993). Honeybees mainly visited recently opened flowers of C. reflexa for pollen and rarely visited older flowers. Recently-opened flowers were in male phase and rich in pollen while older flowers were in female phase. Because they preferred male flowers, honeybees only pollinated the occasional Correa flower and were not as effective as birds which visited all floral stages. When both birds and honeybees had access to the flowers, 26. 1% of flowers produced fruit. However, when birds but not honeybees were excluded from flowers fruit production dropped significantly to 10.7% (Paton 1993). Thus honeybees were not as effective as honeyeaters in pollinating this plant.

Visits to flowers by honeybees often outnumbered those by birds (eg table 6; Paton unpubl.) and honeybees were often the first to visit recently-opened flowers, dislodging 87% of the pollen on their first visit (Paton 1993). In comparison, several species of honeyeater only dislodged an average of 34-53% of the pollen on their first visit to a virgin flower (Paton 1991, unpubl.). On many occasions recentlyopened flowers 41 were visited several times by honeybees so little pollen remained when the flower was first visited by a bird. At times honeybees even chewed undehisced anthers to rob pollen from flowers that were just opening.

Paton (1993) measured the impact of this pollen loss on subsequent pollination by birds with a series of simple aviary trials. Captive Eastern Spinebills, Acanthorhynchus tenuirostris were presented with eleven C. reflexa flowers: a source flower that supplied pollen and ten sink flowers that received pollen. Sink flowers had been emasculated before the anthers dehisced and so contained no pollen. Thus any pollen that these flowers received during trials must have come from the source flower. The ratio of source flowers to sink flowers in these trials approximates the natural ratio. The flowers of C. reflexa live for about nine days. Consequently one flower in nine would be expected to have released pollen in the last 24 hours. In each trial, captive honeyeaters were allowed to visit each of the flowers between five and ten times (table 18), similar to the frequency with which birds visited Correa flowers in the field (eg table 6). All flowers were then retrieved and the pollen deposited on the stigma of each flower counted under a microscope. The amount of pollen initially present at the source flower was then varied and any differences in pollen receipt by the sink flowers measured. To vary the amount of pollen at source flowers and mimic pollen loss due to bee visits, up to seven of the eight anthers were removed. When pollen was removed from source flowers, significantly fewer sink flowers received pollen (table 18) and the total number of pollen grains landing on their stigmas was also significantly reduced (table 18). Appropriate field work is now required to determine if this also happens in the field and leads to reduced production of seeds for this plant in areas stocked with honeybees, but the data indicate the potential for honeybees to have a negative effect on pollination rates for this bird-pollinated plant.

Pyke (1990) reports a similar finding for Christmas Bells, Blandfordia nobilis flowering in Barren Ground Nature Reserve near Jamberoo, NSW. In some years, honeybees apparently removed so much pollen that the effectiveness of pollen transfer by honeyeaters, the native pollinators, was reduced and seed production consequently reduced but Pyke (1990) provided no quantitative data.

These studies showing the negative effects of honeybees on the seed production of plants need to be balanced against other studies where honeybees have been found to aid rates of pollination.

In Ngarkat CP seed production by B. ornata was significantly enhanced at sites stocked with honeybees in each of three years (table 19). At sites without honeybees seed production ranged from 4-7 seeds per inflorescence, while at sites with honeybees seed production ranged from 8-11 seeds per inflorescence. There were also differences between years in overall seed production but in all years seed production was significantly increased in the presence of honeybees.

Table 17
Fruit production by flowers of Callistemon rugulosus placed inside and outside wire mesh cages at three levels of honeybee activity: low (ca 5 bees/1000 flowers daily maximum); medium (ca 10 bees/] 000 flowers); and high (ca 15 bees/1000 flowers). The wire mesh cages excluded birds but did not alter visitation rates to flowers by honeybees. Note that when honeybee activity is high bird activity is low (table 11). Data were collected from 12 plants with each plant having a caged and uncaged treatment, and then pooled. Total number of flowers in each treatment are given in parentheses. Although there was significant heterogeneity between replicates, that heterogeneity results in analyses of pooled data being, if anything, conservative. Analyses of pooled data show: that fruit production increased significantly for caged flowers as honeybee activity increased (X2 = 45.1, df = 2, p<0.001); that fruit production at caged flowers was significantly lower than at uncaged flowers (X2 = 181.6, 121.6, and 12.8 for low, medium and high densities of honeybees respectively df = 1, p<0.001); and that uncaged flowers exposed to birds .and bees declined significantly with increases in honeybee activity (X2 = 38.0, df = 2, p<0.001).
  % flowers setting fruit inside flowers setting fruit outside
Level of honeybee activity
wire cages (honeybees only) % wire cages (honeybees & birds)
6.7 (735)
35.1 (770)
15.3 (2584)
27.9 (2662)
17.1 (1317)
22.6 (1330)
Table 18
Influence of pollen availability at source flowers on subsequent dispersal of pollen to sink flowers by captive Eastern Spinebills visiting Correa reflexa flowers. Differences in the number of sink flowers receiving pollen and in the number of pollen grains deposited were significant (ANOVA, F = 7.9, df = 2,37, p <0.001 in both cases). Note that source flowers with 1 intact anther (instead of 8) have had 87% of their pollen removed, which is the average quantity removed by honeybees on their first visit to a flower. This treatment mimics the removal of pollen by honeybees in the field (see table 7) where honeybees account for up to 93% of the pollen produced by the plant at times. With 87% of the pollen removed only 25% of the sink flowers are pollinated by Eastern Spinebills and those flowers only receive an average of 5.8 pollen grains each. More flowers are pollinated and the quantities of pollen deposited are significantly higher when pollen is available. Values given are means + se.
Intact anthers on source No. of trials No. of probes into sink flowers No. of sink flowers receiving pollen Total no. of grains deposited on the ten sink flowers
8 15 7.3 ± 1.0 6.2 ± 0.7 89.6 ± 14.0
4 12 7.9 ± 1.0 5.4 ± 0.7 51.8 ± 19.1
1 13 6.9 ± 0.7 2.5 ± 0.7 14.6 ± 4.5
Table 19
Seed production for Banksia ornata at sites stocked and not stocked with honeybees at Ngarkat CP in three separate years. Seed production was consistently and significantly higher at sites stocked with honeybees in each of the three years. At sites not stocked with honeybees seed production was enhanced significantly following additional cross- pollination by hand, but was not at sites that were stocked with honeybees. In 1990 inflorescences only received a single supplementary pollination but in subsequent years each inflorescence was given supplementary cross-pollination at least twice. This accounts for the lower seed production following additional pollination in 1990, since on any one day only a percentage of the flowers on an inflorescence would be receptive, and so fewer flowers received additional cross-pollination in that year. Data show the mean ± se (number of follicles (equivalent to seeds) produced/ inflorescence). For convenience in presentation, data from replicate sites in each year have been pooled. In 1990 and 1992 five of fifteen sites were stocked with commercial loads of honeybees (40-100 hives), while in 1993 nine sites received honeybees. Sample sizes in 1990 were reduced because 1 1 of the 15 sites were burnt in a wildfire before follicle production could be scored. Drought conditions prevented hives from being placed on any sites in 1991.The table only includes data collected on inflorescences close to the central point of each site since this was where the greatest reduction of floral resources (by honeybees) took place.
  Number of follicles produced per inflorescence
natural rates rates following additional cross pollination
Sites without honeybees
3.76 ± 0.26(181)
6.08 ± 0.37(91)*
5.43 ± 0.54 (500)
9.21 ± 0.58 (500)*
7.32 ± 0.64 (225)
11.47 ± 0.24 (225)*
Sites with honeybees
11.30 ± 0.60 (127)**
10.62 ± 0.69(71)
8.30 ± 1.04 (250)**
9.81 ± 0.87 (250)
10.92 ± 0.54 (252)
12.27 ± 0.59 (249)
* significant increase in seed production following additional cross pollination (p<0.001)
** significantly higher seed production at sites stocked with honeybees (p<0.001)


Although these figures show that honeybees enhanced the seed production of B.ornata they do not and cannot be used to infer that honeybees were better pollinators than native fauna. The primary reason for the poor performance by B.ornata in Ngarkat CP was insufficient native fauna, not ineffective pollination by those that were present.

In this reserve native pollinators were unable to fully exploit all of the floral resources being produced by B.ornata (see above) and unable to pollinate sufficient flowers to maximise seed production. At sites not stocked with honeybees, seed production was easily enhanced following additional crosspollination, indicating that native pollinators were insufficient. At sites stocked with honeybees additional cross-pollination of flowers failed to enhance seed production any further (table 19). Thus honeybees were contributing significantly to the production of seeds by this plant and their activity at flowers was sufficient to guarantee that a full complement of seeds was produced.

Although these data indicate that B. ornata benefits from the presence of honeybees in Ngarkat CP, increased seed production may lead to B. ornata becoming more prominent in those heathland communities in future generations, to the detriment of other plants.

Enhanced seed production has also been reported for a few species of eucalypts following the introduction of beehives into areas (Loneragan 1979; Moncur and Kleinschmidt 1992; Moncur et al. 1993; Moncur pers. comm. ). In most cases this enhancement occurred in stands of eucalypts where natural rates of pollination were low, including eucalypt seed orchards and occasionally natural stands (Moncur and Kleinschmidt 1992, Moncur et al. 1993; Moncur pers.comm.). In each of these studies seed production was measured only as the number of viable seeds produced per capsule and no data were given on the numbers of flowers that succeeded in setting capsules. Furthermore all the studies lacked replication and did not control for possible locational and seasonal effects on seed production. Conclusions that honeybees have enhanced the seed production for these eucalypts, therefore, may be premature.

Comparisons with overseas research on competitive interactions between honeybees and native flowervisiting fauna and flora

Honeybees have also been introduced to the American continent where both descriptive and experimental studies have investigated the effects that honeybees (particularly Africanized honeybees) may be having on native bees. These studies not only provide further examples of potential interactions between introduced honeybees and native fauna, but they also highlight the complexity of the interactions and the difficulty in assessing impacts. Most of the experimental work has been conducted by David Roubik in Central America, but several North American and German studies have also been conducted.

Schaffer et al. (1979) recorded ari inverse relationship between the numbers of honeybees and other native bees (Bombus, Xylocopa) working patches of flowering Agave schottii in Arizona, with honeybees predominating in the most productive patches. There was also some temporal separation in the activities of these bees, with honeybees being more active during the first few hours of the day when standing crops of nectar were highest. Schaffer et al. (1979) suggested that this reduced the quantities of nectar available for other animals and hence the use of Agave schottii by native bees. Ginsberg (1983) studying the foraging of Apis honeybees and native bees on a variety of wildflowers near Ithaca, New York found a similar pattern with honeybees dominating the richer sources and native bees being more prominent at poorer sources. Schaffer et al. (1983) subsequently manipulated the availability of nectar at stalks of Agave schottii by excluding ants. Ants foraged on nectar both during the day and overnight, and consumed a substantial share of nocturnal nectar production. In the first manipulation, ants were excluded with tanglefoot from 10 of about 130 flower stalks in a 1 ha area. These stalks were then visited by greater numbers of both honeybees and bumblebees, Bombus compared to control stalks where ants still had access. Ants were then excluded from all of the stalks in and around the study area. Following this manipulation the numbers of honeybees increased again, as did the numbers of small solitary bees but the numbers of Bombus did not increase.

During these experiments, two hives of Cordovan honeybees were present at the site. The subsequent introduction of two more hives had no discernible effect on the numbers of honeybees working the flowers. These hives were then removed and for a short period (3-5 days) the numbers of honeybees working Agave schottii were low. Over the next 3-5 days the numbers of Bombus, small solitary bees and feral honeybees (distinguished from hive bees by colour) all increased, but once feral honeybee densities reached those present before the hives were removed, the numbers of bumblebees and small solitary bees declined. 44 Three important findings come from this work.

  1. Manipulations of honeybee numbers by shifting small numbers of hives into and out of areas may not lead to any significant change in the numbers of honeybees working the flowers within study plots.
  2. The behavioural responses of native fauna may vary depending on the scale of the manipulations. At one scale a behavioural response for Bombus was detected but not for smaller solitary bees while at a larger scale no response was detected for Bombus (following one of the manipulations) but was detected for smaller solitary bees. Thus in designing field experiments to test the impact of honeybees on native fauna the scale and direction of the manipulations must be carefully considered.
  3. Competition for floral resources may still exist even when competing taxa forage at different times of the day and in this case the consumption of nectar by ants, particularly overnight, reduced the amounts available for bees during the day and the use of Agave patches by bees.

Roubik's work in Panama and French Guiana has carried these experimental studies much further. He considers not only behavioural responses of stingless social bees to manipulations of honeybee densities and food resources but also measures the effects of introductions of honeybees on colony performance of these native stingless bees and longer term changes in the abundances of native bees in general. Most of the experimental manipulations were conducted before 1983 in areas where feral Africanized honeybees were either absent or scarce. The manipulations involved exposing native bees to 5-22 hives of honeybees for short periods of time, ranging from 1-30 consecutive days. These experimental studies have shown that honeybees non-aggressively displaced some native social bees from flowers and artificial feeders and that colonies of honeybees were superior to native social bees in their ability to locate and harvest rich floral resources and to respond to changes in the availability of floral resources (Roubik 1978, 1980, 1981, 1982a). Despite being displaced from some floral resources there were no detectable changes in the amount of food stored or brood produced by colonies of native bees during exposures of 30 days (Roubik 1982a, 1983). These native bees, although overlapping extensively with honeybees in diet, were generalists and Roubik suggests that they simply switched to other floral resources and so avoided serious competition. However, changes in colony performance may not become evident within 30 days or the changes may not be of sufficient magnitude to be detected within that time period. Furthermore, Roubik (1982a, 1983) suggested that the actual density of honeybee colonies used in the experiments (if their foraging range was taken into account) was probably equivalent to a density of about 1 honeybee col/km² and that this density may not be sufficient to elicit a significant response or properly test the impact of colonising Africanized honeybees on native social bees. Based on estimated densities for Africanized honeybees in other parts of South and Central America, densities approaching 10 feral col/km² might eventually establish.

Subsequently Roubik et at. (1986) examined the foraging activity and resource harvests of 17 colonies of 12 species of native bees in the presence and absence of 20 colonies of honeybees. When honeybees were present the foraging activity for most colonies of native bees declined and for 7 of 31 cases the decline was significant. Colonies of native bees had rare and brief periods of intensive harvesting. During these peak periods up to 51% of the food being harvested by a colony was being collected in just 4% of the time that colonies were active (Roubik et at. 1986). For all colonies of native bees these peaks of activity were diminished when honeybees were present and as a consequence the amount of food harvested by colonies was reduced by about 25% (Roubik 1988). Roubik et al. (1986) estimated that at this rate some native bees may disappear within 10 years.

Longer-term studies have shown that there have been gradual increases in the proportional abundances of honeybees and decreases in native bees at flowers in Panama and French Guiana (Roubik 1988, 1991, unpubl.). In 1977, one year after honeybee arrival, honeybees accounted for 7% of bees at flowers but in 1981 and 1982 they accounted for 67% and 75% respectively (Roubik 1988). Despite this there was no abrupt drop in the numbers of native bees counted at baiting stations or caught in traps 45 during the -first few years after Africanized honeybees arrived in the area but careful analysis of individual species or sets of species were still to be undertaken (Wolda and Roubik 1986; Roubik and Ackerman 1987; Roubik 1988).

Again, these studies by Roubik highlight the difficulty of conducting effective field experiments that test for impacts of honeybees on native bees. Simply introducing a few colonies of honeybees to an area for short periods of time may not be sufficient to elicit a measurable response at a population level. Various behavioural strategies may allow native bees to cope with short-term or localised food losses. First, most stingless social bees in Central America forage at a wide range of plants (eg Roubik et al.1986) and can probably shift to othet resources if displaced from particular plant species by honeybees. On Barro Colorado Island, Panama, Roubik and Aluja (1983) showed that individuals of two species of stingless social bee were capable of navigating back to their colonies over distances of 1.5-2.1 km. Thus stingless social bees from individual colonies may be able to exploit floral resources from areas of 7-12 km², providing considerable scope for avoiding areas and plant species being used extensively by honeybees. Second, these bees often store surplus resources well above their immediate needs and so can wait out periods of resource scarcity by using these reserves (Roubik 1982b) without changes in abundance taking place. Furthermore, average worker life spans may be longer during periods of food shortage (possibly because the bees are less active) reducing the rates of brood production needed to maintain colony size and presumably reducing demands on food stores (Roubik 1982b). Australian native bees may also have an ability to fast and so wait out periods of food scarcity and/or inclement weather (eg Sugden 1988).

In Germany, Evertz (1993) has examined interactions between honeybees and other species of wild bee. Three separate studies - were reported. In one study natural bee communities and population sizes of selected species were monitored at four sites over three years. Colonies of honeybees were introduced to two of the sites in the second year and the abundances of wild bees, particularly oligolectic species (eg Andrena vaga and Colletes succinctus), were reduced along line transects in those areas with added hives of honeybees. Evertz (1993) suggests the species most affected were those whose pollen sources were harvested extensively by honeybees. In the second study, Evertz showed that the numbers of Colletes succinctus found in meadows increased with distance from an apiary. In the final study, Evertz (1993) introduced nesting blocks of 250 coccoons of leaf-cutter bees (Megachile rotunda) to areas of lucerne planted on land being reclaimed after coal mining. The coccoons subsequently hatched and Evertz then scored the number of new coccoons produced two months later in areas with and without added colonies of honeybees. In two of four separate experiments the number of new coccoons produced in areas without honeybees was twice those produced in areas with honeybees, in another it was five times higher and in the fourth there was no difference. Evertz (1993) attributed the variability in competitive response to differences in the availability of floral resources during the different experiments with competition being greater when floral resources were more limited.

Information on interactions between honeybees and other fauna and flora is scant; but two studies support some of the work conducted in Australia. Wilson and Thomson (1991) showed that extensive loss of pollen to introduced Apis mellifera (and Dialictus rohweri) reduced the quantities of pollen being deposited at the stigmas of Impatiens capensis. However, they did not consider whether this reduced rate of pollination led to reduced seed production. Roubik (unpubl., pers. comm.), however, has recently found that seed production for a native South American legume was reduced when feral honeybees displaced native bees from flowering patches, the displacement being most severe in slightly disturbed habitats. The results of these studies are consistent with those reported for Callistemon rugulosus and Correa reflexa in Australia (Paton 1993; and above). Other studies have also shown that nectar-robbing bees (not Apis) can reduce the frequency with which nectarfeeding hummingbirds visit flowers and lead to reduced seed production as well (eg Roubik 1982c; Gill et al.1982), illustrating the potential for nectar-robbing bees to displace nectar-feeding birds and disrupt pollination processes.- 46

Competition between feral honeybees and hollow-frequenting native fauna

Possums, gliders, bats, some dasyurids and a wide variety of birds (eg parrots, cockatoos, some ducks, some falcons, owls, owletnightjars, kookaburras, kingfishers, tree martins, treecreepers) and some reptiles may use hollows in trees for roosting and nesting in Australia (Saunders 1979; Saunders et al. 1982; Pruett-Jones et al. 1980; Tidemann and Flavel 1987; Lunney et al. 1988; Lindenmayer et al. 1990a; Joseph et al. 1991; Mawson and Long 1994). Some of these species could potentially compete with feral honeybees for hollows. However, there is no strong evidence of competition between feral honeybees and any of these hollowfrequenting native fauna.

Concern about competitive interactions is based largely on recorded instances of honeybees displacing cockatoos from hollows (Saunders 1979; Matthews 1984; Bell 1987; Rowley 1990) or of honeybees occupying hollows that had previously been used by bats, parrots or owletnightjars (Tidemann and Flavel 1987; Mawson and Long 1994; McDonald 1994). These displacements, however, may only involve a small proportion of the population and have no significant effect on the population sizes of native fauna, particularly if other hollows are available for use. For example, Rowley (1990) records only two incidences of feral honeybees displacing Galahs from nesting hollows out of some 602 nesting attempts. Similarly Saunders (1979) recorded only two cases of honeybees displacing White-tailed Black Cockatoos from their nests in over 300 nesting attempts that were recorded over six or seven years. Saunders still regarded honeybees as being a problem perhaps because they occupied other hollows that were not used by the cockatoos.

Others have attempted to assess the potential for competitive interactions by estimating the likely overlap in dimensions of hollows used by selected native fauna and honeybees and/or by considering the occupancy rates for hollows in an area (Stace 1988; Wapshere 1988; Manning 1993a; Oldroyd et al. 1994).

Information on the sizes of hollows used by Australian native fauna are scant except for some information on the dimensions of nest cavities used by several cockatoos and parrots in Western Australia (Saunders et al.1982), bats in south-eastern Australia (Tidemann and Flavel 1987) and information on the sizes of artificial nest boxes used by various mammals and birds (eg Menkhorst 1984). Unfortunately there are no data on the sizes of natural cavities used by feral honeybees in Australia and so the comparisons that have been made (Wapshere 1988, Stace 1988) have used information collected on feral honeybees in other countries.

Overseas studies have reported that feral colonies of honeybees use cavities of 10-450 L capacity though most were in the range of 20- 80 L (Seeley and Morse 1976, 1978; Seeley 1977; Jaycox and Parise 1980; Rinderer et al. 1981,1982; Winston 1987; Schneider and Blyther 1988). Most of this information was based on just 21 natural nest cavities found near Ithaca, NY that Seeley and Morse (1976) cut open and measured and a further 28 nests in manmade structures that Seeley (1977) measured, although Schneider and Blyther (1988) provide comparable data for an African race of the honeybee Apis mellifera scutella. Other studies merely recorded the preferences of swarms when offered a limited range of different-sized cavities. These showed that swarms would use cavities that ranged in size from as little as 10 L to at least 100 L. Volumes of nest cavities for feral honeybees from overseas overlap with the volumes of nest cavities used by various cockatoos (Saunders et al. 1982) and nest boxes used by a range of possums, gliders and birds in Australia (Menkhorst 1984; McDonald 1994) but other than indicating the presence of an overlap nothing can be concluded about potential competition from these comparisons.

Although most studies have not thoroughly examined all hollows the consensus is that natural hollows in most areas are underused. Saunders (1979) and Saunders et al. (1982) recorded rates of occupation for hollows suitably sized for various cockatoos and indicated that 29-53% of these hollows at two sites were occupied by birds (of eight species) in spring over a number of years. These data suggested that hollows were not in short supply within these breeding areas and that occupation of some hollows by honeybees would not have affected these 47 birds (though habitats with suitable hollowcontaining trees may have been limiting). Braithwaite et al.(1984) examined 104 hollows in trees felled in the Eden area, NSW and reported that animals used 23 of the hollows and honeybees four, again suggesting that hollows were not in short supply. Gates (1992) recorded occupancy rates for hollows in 86 dead Eucalyptus camaldulensis (64) and E. largiflorens (22) at Disher Creek near Renmark, South Australia. Only 61 (13%) of the 458 hollows were being used for nesting by birds and a few (< 10) were occupied by bats. No honeybees were recorded. In remnant woodland in the southern Mt Lofty Ranges, South Australia, Paton and coworkers (unpubl.) examined 511 trees and recorded 137 hollows in 73 of the trees. Only one of the hollows (<1%) was occupied by honeybees. Lindenmayer et al.(1990a) reported that 31% of 1125 hollow trees stagwatched in mountain ash forest of the Central Highlands, Victoria were occupied by arboreal mammals but occupation rates varied from 0% to 100% of hollow trees from site to site. Many also housed bats but details were not recorded and no information was provided on the use of these trees by birds or honeybees. These forests, however, are generally regarded as being poor sites for feral honeybees (see table 2). Finally, Oldroyd et al.(1994) found that only 0. 7% of possible hollows and 1.3% of trees examined at Wyperfeld NP, Victoria were occupied by honeybees. No attempt was made to determine what proportion of these hollows were being used by native fauna except that Regent Parrots, Polytelis anthopeplus used 0. 4% of the possible hollows during spring. Oldroyd et al.(1994) carried the analyses a little further and concluded that 52% of the feral honeybee colonies in the area were using hollows that either had entrances that were smaller than those being used by Regent Parrots or were closer to the ground and suggested that this would help to reduce any competition between these two taxa. Burbidge (1985) concluded that Regent Parrots were more likely to be limited by food supply rather than availability of hollows.

These conclusions contrast with observations made by Frank Noelker of a reduction in Regent Parrots at Lake Albacutta coinciding with increasing feral bee occupancy of redgum hollows (R Begg pers. comm.). Over a number of years the percentage of hollows occupied by honeybees increased to 16% and over the same period 12 pairs of Regent Parrots abandoned the area. Although these observations suggest that competition for hollows between feral honeybees and some hollow-nesting fauna may exist in some areas, other factors such as changes in food supply and disease could have eliminated birds from an area.

The generally low rates of hollow occupancy by honeybees (typically < 1% of hollows and <1% of trees) reported for woodland and forested areas, however, may not be maintained in agricultural areas and other areas where the number of hollows is low. For example, Paton and Eldridge (unpubl.) examined 416 remnant eucalypts along roadways and in agricultural areas of the South East of South Australia. Ninety-three trees had hollows and eight of these had feral colonies of honeybees. Although only 2% of all the trees examined in this area had feral colonies, 9% of the trees with hollows had feral colonies. Losses of further trees, particularly the larger trees that generally contain hollows, may increase occupancy rates for feral honeybees and result in competitive interactions between honeybees and native wildlife for the few remaining hollows.

Most studies on hollow availability have simply recorded the numbers of trees above a certain diameter that contain hollows, the numbers of entrances that are present in the trees and whether hollows are being used during a short interval of time (usually a few minutes of observation time or during a brief inspection of the hollow; eg Braithwaite et al.1984; Oldroyd et al., 1994). The sizes of the hollows behind these entrances are rarely measured or assessed to see if they are potentially usable by different taxa. Hollow limbs and trunks may be too large, too small, lack an adequate floor and or be used-by other animals (eg bats) that cannot be determined without internal examination.

The choice of hollows for use might also be influenced by ease of access for possible predators (eg goannas, snakes) and availability of nearby perches that facilitate access for users (eg Tidemann and Flavel 1987; Smith and Lindenmayer 1988). Some hollows are also frequently used by different species of birds and mammals at different times in the year or in different years 48 (Saunders 1979; Menkhorst 1984; McDonald 1994) and previous use may influence subsequent use. Native fauna might also regularly avoid certain hollows and or shift hollows as part of anti-predator, anti-disease or some other behaviour (Tidemann and Flavel 1987; Lunney et al.1988; Taylor and Savva 1988) or exclude other species or conspecifics from an area so not all hollows are occupied at any one time (Saunders et al.1982; Menkhorst 1984; Smith and Lindenmayer 1988; Lindenmayer et al.1990a). Thus a network of suitable hollows may be required by a species within an area and, although only a small proportion of them may be occupied at any one time, over a longer time period a much higher proportion of the hollows may actually be used by native fauna. Studies reporting that only a small proportion of hollows are occupied at any one time often fail to consider these possibilities and have largely assumed that suitable unoccupied hollows are surplus to wildlife requirements. This may not be the case.

Assessments of potential competitive interactions between feral colonies of honeybees and native hollow-dependent fauna are far from satisfying. Honeybees, however, occupy only a small proportion of the hollows and for this reason they are not considered to be a major problem in many of the forested areas of Australia. In fact, honeybees are not mentioned in a number of studies on hollows. Of greater concern is the continuing loss of old hollow-bearing trees due to logging of forests, clearing for agriculture and natural decay without replacement (Saunders 1979; Saunders et al.1982; Smith and Lindenmayer 1988; Lunney et al.1988; Lindenmayer et al. 1990a,b; Joseph et al.1991; Bennett et al. 1994; Nelson and Morris 1994; Mawson and Long 1994; Gibbons and Lindenmayer 1995). There are no overseas studies that have examined possible competitive interactions between feral colonies of honeybees and other hollow-frequenting fauna.

Summary and general discussion

Honeybees interact with a wide variety of Australian plants and animals, with records of honeybees working the flowers of at least 200 Australian plant genera. For many plants, honeybees are now the most frequent floral visitors, often consuming more than half of a plant's floral resources. As such honeybees interact significantly with the Australian biota and these interactions need urgent assessment.

A few studies have attempted to measure the impacts of honeybees on native fauna and flora. Studies on native bees suggest their abundance at particular flowering plants is reduced when honeybees are working the flowers but data presented to support this are equivocal. Furthermore studies on reproductive parameters of several species of Australian native bees have so far failed to demonstrate a conspicuous and consistent negative effect. However, these studies may have failed to manipulate honeybee densities adequately to cause a measurable response and second order interactions involving responses by predators or parasites may have confounded the response of native bees. Future studies will need to pay greater attention to the spatial and temporal scales of any experimental studies, and to the type (addition versus removal of honeybees) and magnitude of any manipulations.

The responses of honeyeaters to introductions of honeybees varied. In areas like Ngarkat CP where there were surplus resources during the winter, numbers of honeyeaters did not decrease following the introduction of commercial loads of honeybees. In this particular case some resources still remained unexploited at the end of the day even at sites stocked with honeybees. In other areas where there were no surplus resources individual honeyeaters often held feeding territories. Territorial honeyeaters responded to losses of nectar to honeybees by increasing the sizes of their feeding territories and adjusting the frequency with which they visited particular flowers. Increases in territory sizes of 30-50% reduced population densities by 30-50%. However, whether losses of up to a half of the birds living in an area are critical to the longterm persistence of these honeyeaters is not known. Future work will need to establish if 49 these localised reductions threaten the longterm viability of these birds. Presumably there have been competitive interactions between the birds and bees for approximately 100 years, although the intensity and frequency of this competition may have increased over the last 30-40 years with continued habitat destruction and degradation, and substantial increases in the numbers of managed colonies of honeybees in Australia. Because the birds have survived these past perturbations, they appear not to be threatened. The impact of competitive interactions, however, may be more complex than just the simple exclusion of part of the honeyeater population. Females appear to be displaced more frequently than males and this may affect population dynamics. Further work is required.

Honeybees also influence the production of seeds by various plants. Their presence reduces seed production and/or rates of pollination for several predominantly birdpollinated plants. Other plants experience enhanced production when honeybees are present. Plants experiencing increases in seed production appear to be those that are pollinator limited, suggesting that native fauna are no longer providing an adequate service. Plant-pollinator systems are potentially vulnerable to perturbations like habitat clearance and degradation (Rathcke and Jules 1993) and honeybees may now be important pollinators of native plants in small remnants where native pollinators are deficient (eg Aizen and Feinsinger 1994). Reported and suspected incidences of pollinator limitation in Australian plants largely involve plants, including Banksia, that flower during winter or spring and are pollinated to some _extent by birds (Paton 1988; Copland and Whelan 1989; Vaughton 1991; Whelan and Goldingay 1986, 1989; Goldingay and Whelan 1990). Frequent observations that floral resources are-more abundant during the winter months than during summer months are consistent with this pattern. Several species of eucalypts may also experience shortages of pollinators particularly in plantations and so benefit from attention by honeybees (eg Loneragan 1979, Moncur et al. 1993). These differing responses by the plants need to be considered before implementing management programs for honeybees.

Feral honeybees use hollows that broadly overlap with those that are used by a wide variety of birds and mammals. Initial studies suggest that honeybees only occupy a small proportion of available hollows (often <1%) and that interactions with hollow-nesting fauna may not be substantial. However, few studies make an adequate assessment of the availability of suitable hollows (including internal characteristics) and-in some locations where hollows are rare, significant competition may occur.